Entry,Category shortcode,Publishing organisations,Size,Authors,Biography,Country,Description,Impact,Context,Tools,Lesson,Description portfolio,Cover letter,Languages,Link 1,Link 2,Link 3,Link 4,Link 5,Link 6,Link 7,Link 8
BBC Shared Data Unit,Portfolio,BBC,Big,N/A,N/A,United Kingdom,N/A,N/A,N/A,N/A,N/A,"In March, we reported how hundreds of sex offenders across the UK had slipped off the radar of police by changing their identity to escape the authorities. 

The research involved interviews and FOI requests to police forces and security agencies. We found dozens of examples of abusers who had changed their names and compiled their pictures into a gallery to visualise how widespread this issue was in our television reporting

Two abuse survivors, Della Wright and Lauren-Eden Sullivan, waived their legal right to anonymity to share their personal experiences as part of our reporting. 

Our research was quoted in a debate in a House of Commons Chamber on the day after publication and received widespread coverage in the national and regional Press. 

Six months after our report, then-Home Secretary Suella Braverman announced a plan to ban sex offenders from changing their name and gender - what’s been called Della’s Law.

In April 2023, we worked with BBC's Newsbeat to report that six out of 10 recently inspected UK rental homes failed to meet a proposed new standard for energy efficiency.

In June 2023, we analysed The Gazette to gain a regional picture of the business insolvencies being experienced across the UK as a result of the pandemic and the cost of living crisis.

In July and November, we teamed up with BBC Wales Investigates to take a deeper look into care in the UK over the course of two exclusive stories. 

We found a system in chaos with parts of the UK experiencing 100-day waits for assessments, patients stuck in hospital for nearly a year because there was nowhere for them to go and more than 13,000 care contracts handed back to councils, largely because private companies could not provide enough staff. 

Our investigation found more than 1,300 people had died in the space of a year while waiting for their care package to start and also looked at how a lack of available care was causing large backlogs in the health system. The data was acquired through more than 500 Freedom of Information requests and by interrogating daily hospital discharge data.
The investigation resulted in the BBC documentary I Want to Go Home as well as bulletins across the BBC network and two online stories in July and November. 

In August, we revealed that a black hole in local authority budgets had continued to grow, prompting fears some will not be able to provide basic services. 

The research involved surveying 190 upper-tier authorities in the UK to find out the extent of the financial difficulties facing town halls.

Our investigation featured across BBC platforms, including the BBC News website, News at Six and Radio 4's Today, as well as BBC Wales and around 20 other BBC radio outlets. The story was also published by over 80 local media such as Manchester Evening News and Wales Online.

We followed up with a second piece in January 2024 exploring levels of Town Hall debt.","The BBC Shared Data Unit has continued to promote and support data journalism at a regional level across the UK, whilst making national and local headlines with a series of public interest data investigations. The team’s collaborative model sees it share its data journalism with more than 1,000 local news outlets across television, radio, print and digital. 

During the course of the year, the team have also led week-long masterclasses in investigative journalism and launched a coding course for journalists. We help to strengthen local news output by giving journalists the skills to interrogate data and tell stories of importance to their communities.
A key part of the team’s editorial remit is to find news stories that are hiding in plain sight, using data journalism techniques to explore the impact of local and national government policies on the lives of individuals. Its reporters take the time and space to work on exclusive investigative journalism, using digital techniques to tell stories that would otherwise go untold.

The SDU believes in making its journalism accessible to as many people as possible, and in being transparent and accountable in its reporting. The majority of our data, methodologies and code are published on our website.",English,https://www.bbc.co.uk/news/uk-64624921,https://www.bbc.co.uk/news/newsbeat-65136313,https://www.bbc.co.uk/news/uk-wales-66260332,https://www.bbc.co.uk/news/uk-england-67125440,https://www.bbc.co.uk/news/uk-66428191,https://www.bbc.co.uk/news/uk-67707156,,
MuckRock,Portfolio,MuckRock,Small,N/A,N/A,United States of America,N/A,N/A,N/A,N/A,N/A,"In 2023, MuckRock’s editorial team worked on data investigations with local newsroom partners across the United States, focusing largely on the environment and public health. 
Our “Atomic Fallout” series, published with The Missouri Independent, a new States Newsroom-affiliated politics-focused outlet, and The Associated Press, was a six-month multi-newsroom investigation that revealed, with the help of newly-declassified federal documents, how the U.S. government downplayed and ignored health risks associated with St. Louis radioactive waste for decades. The investigation sparked bipartisan federal legislation that would expand a program compensating those who lived near the Superfund sites, to include Missouri and other states.
Within days, Sen. Josh Hawley, a Republican, and Rep. Cori Bush, a Democrat, pledged action, calling the investigation “devastating” and decrying the federal government’s “negligence."" ""This report confirms what we in the community have known for decades: that for the past 75 years, the federal government actively and knowingly treated St. Louis as a dumping ground for harmful and toxic radioactive waste,"" Bush said. 
“The Air We Breathe,” with the bilingual community newsroom, the Cicero Independiente, is an ongoing yearlong project involving air quality sensors we installed on volunteers’ homes. Those sensors found comparably, and dangerously, high concentrations of particulate matter 2.5, or soot. We hosted a local event in Cicero in August, sharing our initial findings on the town’s air pollution with residents, before publishing a two-part series in November and December. Following the series, Cicero’s town board president publicly criticized the Independiente’s journalists, and threatened to bar their reporters from speaking during public meetings.
“Smoke, Screened,” an investigative series by NPR’s The California Newsroom, MuckRock and the Guardian, focused on the large environmental impact of a little-noticed deregulatory tool found in the federal Clean Air Act. The legal provision has allowed the Environmental Protection Agency to strike pollution from clean air tallies in more than 70 U.S. counties, enabling local regulators to claim the air was cleaner than it really was for more than 21 million Americans. The loophole allows regulators to forgive pollution, and avoid costly cleanup work, caused by “natural” or “uncontrollable” events, including wildfires. Following the series, our reporters presented their work at the annual American Geophysical Union conference for a forthcoming paper.

MuckRock also supported newsroom data projects in other ways. We compiled and released a data repository to track U.S. maternal mortality rates to share with local newsrooms across the country, resulting in several stories in North Carolina, Michigan and Georgia. We continued our investigative series on the U.S. child care industry, called “Disappearing Day Care,” which showed that despite hundreds of millions in federal pandemic relief money pouring into the state, child care facilities are facing huge staffing shortages and parents are struggling with long waitlists for care. And a government spending and data investigation from STAT News and MuckRock shows that the nearly $1.2 billion in taxpayer money for long Covid research has resulted in few tangible results.","[MuckRock](https://www.muckrock.com/) is a nonprofit, collaborative news platform that brings together journalists, researchers and the public to request, analyze and share primary source data and documents in the public interest. In 2021, MuckRock created an editorial unit, which operates as a hybrid newsroom and journalism support organization. 
The newsroom’s editorial policy defines its work as “both collaborative and independent reporting efforts on issues of public importance.” The editorial team works with existing local publications that have large data, analysis and document needs, supplementing both the specific partner investigations while helping them build their own capacity for reporting in the future. Through these partnerships, MuckRock works alongside other, early-career journalists to provide much-needed help in obtaining and analyzing primary source documents and data for investigative projects.
Each year, we manage dozens of partnerships with local newsrooms around the country, helping these newsrooms increase their impact while covering critical stories that would otherwise go untold. There are few newsrooms that have a stated mission as a local journalism support organization and even fewer with federal, state and local open-records and data expertise.
Our work has grown to more than a dozen partnered investigations with local newsrooms across the country, including deregulation and wildfire management in California; sex trafficking and migrant issues in Texas; nuclear waste and its resulting public health effects in Missouri; and air pollution with our own sensor journalism network in Cook County, Illinois.","English, Spanish",https://www.muckrock.com/project/atomic-fallout-records-reveal-the-federal-government-downplayed-ignored-health-risks-1164/,https://www.muckrock.com/project/smoke-screened-the-clean-air-acts-dirty-secret-1117/,https://www.muckrock.com/project/the-air-we-breathe-how-industry-is-polluting-ciceros-air-1168/,https://www.muckrock.com/project/disappearing-day-care-child-care-crisis-in-the-us-1161/,https://www.muckrock.com/news/archives/2023/apr/20/long-covid-nih-billion/,https://www.muckrock.com/news/archives/2023/jul/25/tracking-maternal-mortality-rates/,,
Narasi,Portfolio,Narasi Newsroom,Small,N/A,N/A,Indonesia,N/A,N/A,N/A,N/A,N/A,"1. This report tells the story of how Russia utilizes a local media outlet, Tribun Timur, to spread propaganda to Youtube audiences in Indonesia. This relationship is symbiotic mutualism, through Tribun Timur Pro-Russian war content can remain on air without being banned and taken down by Youtube. For Tribun Timur, Russia's algorithm manipulation efforts were able to make them gain many views which were converted into ads from YouTube reaching millions of dollars. 

This analysis reveals how YouTube's algorithms are manipulated to promote certain content, creating a misguided reality for viewers.
Nearly 60% of the Tribun's videos were war-related, with comments dominated by the Russian language, suggesting a foreign influence in the narrative presented.

Further investigation indicated the involvement of Russian web forces in this operation, with evidence pointing to St. Petersburg, Russia. We experimented with creating pro-Russian content on Youtube and distributing it to telegrams with a tracking script.

In-depth experiments and analysis revealed how Indonesian-language Ukrainian war content attracted audiences in Russia, including the alleged involvement of Russian tourists in Bali in this propaganda operation.

In conclusion, the case of Tribun Timur illustrates how local media can be caught up in global propaganda networks, with Russia as the main actor in manipulating digital algorithms to influence public opinion. 

2. General Santos City, is one of the most important fishing centers in Asia-Pacific. The problem is that the majority of tuna in Gensan comes from Indonesia.

When Susi Pudjiastuti was the Minister of Fisheries, she made a policy to eradicate illegal fishing which caused the supply of fish to Gensan to decrease drastically. 

However, in 2019, after Susi Pudjiastuti was no longer , illegal fishing activities by Filipino vessels from Gensan started to increase again in the border area. 

We used AI technology to detect the activities of vessels that docked at Gensan harbor during 2015-2020. We also used data on the lights shining in the Sulawesi Sea to strengthen our conclusions. In fact, fish thieves from the Philippines came back to steal fish after Jokowi's policy softened in the second term of his presidency.

3. Royal Golden Eagle Group (RGE) and APRIL, major players in the wood pulp industry in 2015 pledged to eliminate deforestation from all its production operations, but field investigations using Airtag Apple and geospatial data analysis show this pledge is not being fulfilled. RGE and APRIL continue to obtain wood from deforestation, despite having committed not to do so.

These greenwashing practices are revealed through the use of shadow companies and shell companies in tax havens, making their ownership structures complex and difficult to understand.

These greenwashing schemes also affect the global fashion industry. Well-known brands such as Marks and Spencer, Esprit, H&M, Uniqlo, and Adidas source viscose rayon from RGE, which is considered environmentally friendly. Ironically, some of these brands have committed to sustainability and zero deforestation.

The ""greenwashing domino"" effect occurs when one company's greenwashing triggers a chain reaction in its business partners, spreading inaccurate information to consumers.","Our editorial staff consists of less than 35 people (small newsroom). Our reporting platform focuses on video reports and our main target audience is young people. Although we are a small newsroom, we have considerable influence on platforms such as Youtube, TikTok and Instagram. Since 2019, the investigation and data division was formed, with me as a leader. Investigative reports in video platforms using data and OSINT in Indonesia are very rare. We are trying to fill that void. In addition to extensive research, a creative approach is needed so that the public understands the complex data. 

Last year, we were nominated in the portfolio category for small newsrooms at the Sigma Awards. Before that, in 2019, our reporter Sharon Sumolang was nominated in the young journalist category. We hope this year we can be nominated and win. Because, in 2023, we experimented and covered big data processing with the latest methods. Among them: installing link traps in telegram accounts, using AI in fish theft cases and installing trackers on barges of deforestation perpetrators. (You can read the details in the portfolio description section.)

What we do is very unusual for media in Indonesia and Southeast Asia, especially non-documentary video-based media. We want to spread new ways of how journalists search for and work with data. For example: combining spatial data of forest imagery with Apple airtag trackers installed on barges, or using ship-identifying AI tools to calculate fishing activity. Since our target is young people, we are also required to process the complex data with a visual approach in the style of young people who are colorful and not rigid. The content can be cut short and shared on platforms such as Tiktok or IG reels without compromising the investigative context of the stories","Indonesia, English",https://www.youtube.com/watch?v=6CPfN5hONM4,https://www.youtube.com/watch?v=r6yU0M8WOM8,https://www.youtube.com/watch?v=qUy8Eh051SU,,,,,
Pulitzer Center,Portfolio,"Armando Info, InfoAmazonia, La Liga Contra el Silencio, Repórter Brasil, the Organized Crime and Corruption Reporting Project, The Intercept Brasil, Mongabay.",Big,N/A,N/A,United States of America,N/A,N/A,N/A,N/A,N/A,"The Rainforest Investigation Network, a community of investigative journalists created by the Pulitzer Center, has supported collaborations that have enhanced environmental coverage. In this portfolio, we present a series of unique stories and groundbreaking data analyses published in 2023 that serve as a blueprint for other journalists and civil society actors investigating criminal activities in the Amazon rainforest.

Our portfolio comprises four projects that used data to reveal how organized crime and other illicit activities converge to destroy the Amazon. In our portfolio, you will find:

Amazon Underworld: a series of nine stories published by Armando Info, InfoAmazonia and La Liga contra el Silencio, written by a team of 37 journalists from 11 countries and led by journalist and RIN Fellow Bram Ebus. This series uncovers the extent of illicit activities that destroy the Amazon. For that, it created a mapping methodology that indicates the presence of armed groups in 69%of the Amazonian municipalities.

The Big Green Laundromat: a series of four stories published by Repórter Brasil, and written by Brazilian journalist Hyury Potter. This series shows how the Amazon rainforest has been used as a “dirty laundromat” for organized crime groups and how they take advantage of the fragile regulations and political connections. 

Señores del Oro (The Gold Lords): a series of four stories by Joseph Poliszuk and María de los Ángeles Ramírez, published by ArmandoInfo and the Organized Crime and Corruption Reporting Project. The stories show how Venezuelan gold is trafficked through its borders to feed a global trafficking network.

Forest Thieves: Who Profits From Federal Lands? is a series of five stories and one video by journalist Fernanda Wenzel, published by The Intercept Brasil and Mongabay. The stories reveal how criminal groups— that include politicians, businessmen and public servants— have invested large amounts of money to destroy public forests in the Brazilian Amazon. 

Journalists faced challenges producing these projects and stories, such as difficulty accessing data and collaborating. But that is not all. Unfortunately, Amazon’s illegal economy is extremely dangerous, and investigating it poses a risk to some of our journalists. However, they were thorough in their research and took necessary safety precautions, enabling them to conduct extensive and successful field reporting trips. 

This portfolio demonstrates the scale of the illegal economy destroying the Amazon. In all the stories submitted, we have seen an effort to consolidate databases that were collected firsthand and to work on a deep analysis of thousands of documents and geospatial information. 

With the help of the RIN team of editors and data experts, the findings of these investigations were presented in clear and eye-catching visualizations, maps and graphs, making the information more accessible to our audiences. Furthermore, despite being different projects based in different parts of the Amazon, all research methodologies have been shared, reflected in the quality of the results.","I write this letter on behalf of an alliance of journalists and media outlets that collaborated to reveal how organized crime and illicit groups are destroying the Amazon rainforest. Their projects and stories were published by Armando Info, InfoAmazonia, The Intercept Brasil, La Liga contra el Silencio, Repórter Brasil and the Organized Crime and Corruption Reporting Project (OCCRP). All collaborations were coordinated by the Pulitzer Center’s Rainforest Investigations Network.
The Rainforest Investigations Network (RIN) was launched in 2020 to harness investigative reporting and cross-border collaboration to expose the intersection of climate change, corruption, and governance in the Amazon, Congo Basin and Southeast Asia. 

Since 2020, this network has promoted and supported collaborative investigations that unveiled some of the causes of environmental destruction in these regions. Also, the Rainforest Investigations Network is driven by making an impact, and we are proud to say that some of the projects and stories we have supported have led to societal transformation. 

As an example of these impacts, the database created by the project Amazon Underworld was presented to authorities attending the high-level meeting called the Amazon Summit, organized by the Brazilian government in August 2023. Since then, the data has been used in several publications from local and international press and reports by NGOs and think tanks.

The portfolio we are submitting represents an important breakthrough. The journalists’ use of data analysis, innovative methodologies, and their commitment and passion for environmental journalism uncovered a crucial and underreported issue in the Amazon: the connection between criminal and illicit groups and deforestation in the Amazon rainforest. This is why we, as a team, believe we should be considered for this prestigious award. 

Thank you for your consideration. 

Gustavo Faleiros

Director of Environmental Investigations for the Pulitzer Center","English, Portuguese, Spanish",https://amazonunderworld.org/,"https://armando.info/oro-de-sangre-por-alimentos/#:~:text=Un%20esquema%20que%20involucra%20a,terminaron%20en%20el%20mercado%20global.",https://www.occrp.org/en/blog/17676-how-venezuelan-gold-is-trafficked-through-brazils-borderlands-to-the-us,https://narcogarimpos.reporterbrasil.org.br/en/,https://www.intercept.com.br/2022/11/14/amazonia-desmatamento-pasto-terras-publicas/,https://www.google.com/url?q=https://news.mongabay.com/2023/04/professional-services-abound-for-amazon-land-grabbers-seeking-legitimacy/&sa=D&source=docs&ust=1705436476142667&usg=AOvVaw11i0Fad8epQOMaZXfdZ90R,https://www.intercept.com.br/2023/04/10/video-por-dentro-da-industria-da-grilagem/,
RTVE,Portfolio,RTVE,Big,N/A,N/A,Spain,N/A,N/A,N/A,N/A,N/A,"El servicio público es la razón de ser de RTVE - la radiotelevisión pública española- en general y de DatosRTVE en particular. Y por eso uno de los temas a los que dedicamos más esfuerzos a lo largo del año es el cambio climático. Nuestra cobertura tiene dos ejes principales. Por un lado, los trackers de distintas temáticas que mantenemos actualizados con periodicidad variable en función de la actualidad y que vamos mejorando año tras año. Esto incluye un tracker de incendios donde al mapa de incendios activos en España y Europa -con datos satelitales que procesamos, analizamos y visualizamos- le sumamos el análisis con gráficos sobre los incedios históricos, las hectáreas quemádas, los grandes incendios y mucha más información de contexto para explicar el tema de la mejor forma forma posible. También un tracker de embalses perenne durante todo el año en el que detallamos la situación de cada embalse, así como los datos de las cuencas hidrográficas, el registro de precipitaciones y los indicadores de sequía; un compendio de datos explicados y contextualizados que permite un seguimiento completo del agua disponible en España. 

El segundo eje es el seguimiento habitual de los fenómenos meteorológicos y el cambio climático que realizamos mediante piezas concretas. Desde el recuento de las noches tropicales el pasado verano hasta el repaso mensual de las temperaturas, con contenidos aún más elaborados como el análisis de las piscinas españolas en un contexto de emergencia por escasez de agua. 

Y a todo ello se suman contenidos adicionales y originales, como una calculadora de olas de calor que utiliza los datos desde que la Agencia Española de Meteorología tiene registros hasta la actualidad y los presenta al usuario de una manera personalizada que muestra, al final, cómo estos fenómenos meteorológicos son cada vez más frecuentes.

Elaborar estos contenidos es un reto constante dentro de la sección. El esfuerzo por acceder a todos los datos disponibles y comunicarlos de la mejor manera posible nos ha hecho avanzar en el uso de nuevas herramientas -por ejemplo, nuevas librerías en R para geolocalizar las coordenadas de los incendios; QGIS para manejar los polígonos de las áreas quemadas; o Sentinel para obtener imágenes satelitales- y nos mantiene alerta, siempre pendientes para mejorar nuestra cobertura.","DatosRTVE se creó durante la pandemia de COVID, fruto de la necesidad. Desde entonces, el proyecto ha seguido creciendo tanto en las habilidades del equipo como en los temas que aborda: salud, pero también medio ambiente, política o cultura. Y dentro de esas áreas, todo tipo de temas, tratando de ofrecer enfoques informativos y novedosos a través de los datos y la visualización. Dentro de la sección hay perfiles diversos, desde periodistas tradicionales que se han reconvertido hasta perfiles más técnicos y especializados. Además, una parte fundamental de nuestro trabajo es la colaboración con otros profesionales de la casa, desde el equipo de Infografía hasta los compañeros de El Tiempo. Como responsable del equipo, en esta ocasión presento una apuesta de la sección a la que dedicamos un porcentaje considerable del tiempo de trabajo de cada año, y en el que creemos firmemente. La cobertura del cambio climático es una de las principales líneas de trabajo de DatosRTVE que seguirá crecienco y mejorando.",Español,https://www.rtve.es/noticias/20230907/incendios-ultima-hora/2436349.shtml,,https://www.rtve.es/noticias/20230914/calculadora-olas-calor-espana/2446505.shtml,https://www.rtve.es/noticias/20230703/medio-millon-piscinas-espana-afectada-sequia/2450800.shtml,https://www.rtve.es/noticias/20230426/mapa-lluvias-cuanto-zona/2441344.shtml,https://www.rtve.es/noticias/20230824/estamos-ante-verano-extremo/2454501.shtml,https://www.rtve.es/noticias/20231201/calido-record-noviembre-veranillo-caluroso/2463710.shtml,https://www.rtve.es/noticias/20230831/incendios-devastadores-verano-2023/2454816.shtml
ZEIT ONLINE Team Data and Visualization,Portfolio,Zeit Online,Big,N/A,N/A,Germany,N/A,N/A,N/A,N/A,N/A,"365 days
On the anniversary of Russia's invasion of Ukraine, ZEIT ONLINE offers a completely new perspective on the war to date with the help of a comprehensive collection, evaluation and visualization of data and satellite images.

My neighborhood, a bubble
There is often talk of an urban-rural divide when it comes to political lines of conflict in Germany. In reality, however, many divisions run right through German cities. Whether west or east, metropolis or medium-sized city - similar patterns emerge time and again. ZEIT ONLINE has remapped the political landscape of the republic.

This is how much hotter the current heat would be in the future
One of the hottest days this year was July 9, 2023. It was not only warm in Germany, but in large parts of Europe, especially in the south. Using exclusive research data from the Alfred Wegener Institute, we show what this heatwave would look like in Germany Europe with and without climate change.

Where the Earth is becoming virtually uninhabitable
Traditionally, humans live in what is known as a climate niche. In these places, for example, the average annual temperature is below 29 degrees. With climate change, not only are global temperatures rising, but the regions where it is simply too hot to live are becoming ever larger, shows our data analysis and interactive graphic.

This is where all the wind turbines could be located
Many Germans are in favor of wind power - until it comes to the question of where the wind turbines should be located. However, the energy transition is not possible without wind energy, or more precisely: the energy transition is not possible without tens of thousands of new wind turbines. For the first time, we have calculated that there is enough space for all the wind turbines that would be needed for the energy transition in Germany. 

What climate future will we live in
Climate change will soon fundamentally alter the earth. But what do these abstract scenarios from the Intergovernmental Panel on Climate Change (IPCC) mean? Are 1.5 degrees or 2.7 degrees really still so far away and what will we, our children and grandchildren experience? Our interactive calculator uses scientific model data to show what kind of future a person can expect

The new city
Around a third of Germans, 27 million people, live in large cities. How have their lives changed in recent decades? How do inequality, migration and high rents shape life in cities? A team from ZEIT ONLINE spent a year and a half collecting all available data from the 80 major German cities, from authorities, statistics offices, data service providers - and from the administrations of individual cities themselves. In some cases, the team came across unpublished data sets, while in others it was able to make data available for the first time that documents the development of recent years. For the first time, it is now possible to see in detail how the major trends of the present have transformed Germany's major cities.","The ""Data and Visualization"" team consists of data journalists, designers and developers who work interdisciplinary on graphics, interactive tools and experimental forms of presentation. They are responsible for the implementation of all data journalism projects on ZEIT ONLINE.

Whether wars, climate change or the energy crisis: the team finds the latest data and the right format for every topic. In addition to current news topics, the team also manages to set its own, such as the focus research ""The new city"" (https://www.zeit.de/die-neue-stadt), which lasted over a year and a half. The team collected all available data from the 80 major German cities and was able to show in detail for the first time how the major trends of the present have transformed Germany's major cities. 

The coronavirus pandemic has made data journalism systemically relevant. Never have figures and their presentation been so important, never have decisions with such great social relevance been made on the basis of key figures, data and maps, and this development has continued seamlessly ever since. Will the energy transition succeed? What does climate change mean for me? Questions like these can be answered with data.

At ZEIT ONLINE, data and visualizations are no longer an accessory, but the focus of reporting. In collaboration with other departments, its own community, as well as authorities in Germany and researchers around the world, the team tries to break new ground in journalism every day.

Whether reliable live dashboards, surprising visualizations or elaborate data analyses. The ""Data and Visualization"" team at ZEIT ONLINE tries to raise the bar in data journalism in German-speaking countries with every publication.",German,https://www.zeit.de/daten-und-visualisierung,https://www.zeit.de/politik/ausland/2023-02/ein-jahr-krieg-ukraine-russland-schauplaetze-chronik-satellitenbilder,https://www.zeit.de/politik/deutschland/2023-09/wahlverhalten-staedte-bundestagswahl-wahlkarte-analyse,https://www.zeit.de/wissen/umwelt/2023-07/klimakrise-erderwaermung-hitzewelle-extremwetter-zukunft,https://www.zeit.de/wissen/umwelt/2023-06/klimawandel-erderwaermung-klima-daten-hitze-unbewohnbar,https://www.zeit.de/wissen/umwelt/2023-04/energiewende-deutschland-windkraft-ausbau-windraeder-flaechen-karte,https://www.zeit.de/wissen/umwelt/2023-03/ipcc-bericht-klimawandel-weltklimarat-zukunft,https://www.zeit.de/die-neue-stadt
【圖解】解放軍如何進逼第一島鏈：台海衝突下牽動的美日軍事布局,Single project,報導者 The Reporter,Small,文字／許詩愷、洪琴宣、柯皓翔、簡毅慧；設計／江世民；數據分析／柯皓翔、簡毅慧；多媒體企劃／洪琴宣；攝影／楊子磊；核稿／李雪莉、張鎮宏；責任編輯／張詩芸、汪彥成、陳思樺,《報導者》是台灣第一個由公益基金會成立的網路媒體。秉持深度、開放、非營利的精神，致力於公共領域調查報導；跨領域團隊以數據分析與視覺化技術輔助報導，提供創新數位內容並提升讀者閱讀體驗。,Taiwan,台海局勢全球矚目，2022年美國眾議院議長裴洛西訪台後更牽動日本、台灣、菲律賓等第一島鏈國家防衛升級。《報導者》整理台灣及日本官方報告、以地圖呈現中國解放軍最新動向及各國因應措施，解析區域局勢發展。,"本報導在社群上獲廣大迴響，臉書至少逾1,500次分享、IG影片觀看數破11萬、文章達12萬次瀏覽。

團隊秉持開放資料精神，公開分析數據於Github，使更多民眾、研究人員近用。

專題影響力也擴散至日本、美國等國際媒體，不僅獲日本媒體《琉球新報》訪談報導，美國《華盛頓郵報》亦邀請合作，將中國戰機與戰艦侵襲台海周邊的軌跡分析轉譯成英文版，並取用數據集後製，與其廣大國際讀者分享。","數據獲取與清洗：共軍動向資料取自台灣及日本政府官方逾400份報告，但檔案及繪製格式不一，增加團隊後製難度。記者也多方梳理美日軍事基地、領海、防空識別區（ADIZ）等數據，背後查證及建置成本高。

創新技術：團隊使用內部開發的「捲動式影片」工具，這讓團隊得以用Google試算表控制影片及字幕，以scrollytelling手法串接分鏡，並在多種載具下實現滿版體驗效果，不僅有利地圖敘事，亦強化讀者沉浸感。","團隊使用Python爬蟲批量下載圖片，並首次結合ChatGPT，快速建立影像辨識及裁切方法。

為自圖片萃取地理資訊，團隊運用QGIS確認投影法及坐標，並參考媒體同業方法論，透過Figma人工描圖建立svg檔案，透過Python將svg進行空間對位、使檔案轉換為geojson。

最後，我們使用After Effect的GEOlayers3插件，將地理資料匯入After Effect，進行動態影像製作。","資料開放程度及品質會影響分析難度與台海議題關注度。《報導者》將方法論及數據公開於Medium及Github，可讓同業快速使用相關數據，研究者也能循此進一步分析。本報導的方法論同樣適用於其他主題，有大量圖片檔案需格式化、向量化、地理圖資化的專案都有應用潛力。

流程上，數據記者建置好地理資料後，透過QGIS進行機型及路徑分析，將結果提供文字記者採訪使用，並由專家深度解讀，是數據驅動報導的實例。",N/A,N/A,中文,https://www.twreporter.org/a/taiwanyuji-first-island-chain-military-movement-multimedia,https://medium.com/twreporter/ai-python%E8%88%87%E6%89%8B%E5%B7%A5%E6%A5%AD-%E6%88%91%E5%80%91%E5%A6%82%E4%BD%95%E8%99%95%E7%90%86%E5%8F%B0%E7%81%A3-%E6%97%A5%E6%9C%AC%E8%B6%85%E9%81%8E400%E5%BC%B5%E5%85%B1%E8%BB%8D%E8%BB%8C%E8%B7%A1%E5%9C%96-13b10f9a1c81,https://github.com/data-reporter/PLA_Path_Database,https://www.washingtonpost.com/world/2023/11/13/china-biden-xi-meeting-apec-taiwan/,,,,
Alone and Exploited,Single project,The New York Times,Big,"Hannah Dreier, Michael Keller, Eli Murray, K.K. Rebecca Lai","Hannah Dreier is a Pulitzer Prize-winning investigative reporter for The New York Times.
Michael H. Keller is a reporter on the investigative team at The New York Times with a background in computational journalism.
Eli Murray is a Pulitzer Prize-winning reporter and graphics editor at The New York Times.
K.K. Rebecca Lai works on interactive graphics The New York Times",United States of America,"This data-based series exposed the astonishing resurgence of child labor in the United States and revealed the profound policy failures that have led to this shadow workforce. Drawing on nearly two years of on-the-ground reporting and hundreds of thousands of rows of data about migrant children obtained through a federal lawsuit, The New York Times revealed a hidden workforce of children who fled impoverished homelands only to be trapped in grinding, dangerous jobs in America.","This work had immediate impact. Two days after the first piece was published, the White House launched a sweeping crackdown. “The New York Times story is heartbreaking and completely unacceptable,” White House press secretary Karine Jean-Pierre said. The Labor Department ramped up child labor enforcement, doubling the fines it collected. HHS substantially increased support for migrant children, including legal services and case management. ICE launched a new docket system for migrant children who may be victims of exploitation.

After Dreier’s article about missed warnings, Congress held oversight hearings with the heads of HHS and the Labor Department. Top Biden advisor Susan Rice resigned her post, one of the highest-level departures of Biden’s administration. 

Congress members from both parties introduced more than a dozen child labor bills. 

The reporting spurred lasting change at the state level, too, with at least six states passing enhanced child labor laws.

Global corporations named by The Times imposed stricter hiring standards for U.S. suppliers and began requiring auditors to visit hundreds of factories unannounced and scrutinize night shifts. 

As New York magazine writer Jonathan Chait said, “Hats off to a reporter for breaking a scandal so big the political system feels compelled to respond immediately.”","To find a hidden workforce of migrant children, we developed a new approach to analyzing federal data.

HHS shelters children when they arrive, but after releasing them to sponsors, doesn’t track them further. New York Times reporter Hannah Dreier’s early reporting suggested that children released to distant relatives and strangers were most likely to be put to work. So she filed multiple FOIA requests with HHS—and eventually sued—to obtain 550,000 rows of ZIP code-level data showing where children had been released to nonparent sponsors. We overlaid this with U.S. Census population density data to pinpoint areas with high concentrations of these children.

Dreier also used court records from PACER and state courts, as well as documents from dozens of FOIA requests to the Department of Labor and state agencies to hunt down other outcomes the government does not track. She built a database of migrant children killed on the job, including a 15-year-old who fell on his first day roofing and a 14-year-old who was hit by a car while delivering food. Another database showed how rarely the government prosecuted child labor trafficking cases. Another detailed the crushed limbs and seared lungs migrant children suffered in workplace injuries.","After months of back and forth, HHS gave us its data in an extremely messy format. 

In one case, HHS provided about six years of data as 70 individual files, one for each month, with merged cells, macros, special characters and other issues that made them not machine readable. Each file required manual cleaning just to ensure that the first row was the header.

Some of these problems were careless errors such as typos, but many were political in nature. For instance, many fields initially contained the term UC, for “Unaccompanied Child.” The Trump administration changed this to UAC, for “Unaccompanied Alien Child.” Biden then changed it back. These changes meant the data had references to UC and UAC scattered throughout in inconsistent ways.

We wrote a Node.js script to analyze and collate the cleaned headers. We used OpenRefine to cluster cell values and manually reviewed aggregate counts to find outliers. We found that thousands of rows with seemingly accurate but incomplete data were actually erroneous by checking column totals against other sources. 

We did our analysis in a few different ways and employed R, Python and mapshaper.org, as well as Node.js.","This project was an exercise in making due with non-ideal data. In one case, we asked HHS for county-level data, but they gave us ZIP codes instead. To turn this into a county-level database, we used GeoCorr from the University of Missouri Census Data Center, which provides equivalency tables based on Census populations. We then used a Node.js program to turn our ZIP code data into a county-level map. From there, we could decide where to deploy for on-the-ground reporting.

After intensive cleaning, the thousands of pages of documents obtained through our FOIA lawsuit allowed us to report on a host of government failures. Among other findings, we were able to report the staggering statistic that HHS could not reach more than 85,000 children just weeks after they were released to sponsors. 

After Dreier wrote a piece explaining her process, congressional staffers, academics, journalists and even Labor Department officials requested access to our database. As part of our commitment to helping others investigate, The Times made this data public.

Newspapers including The Boston Globe and The Chicago Tribune launched their own investigations, and other outlets, including USA Today, have more data-based migrant child labor reporting in the works.",N/A,N/A,English,https://www.nytimes.com/2023/02/25/us/unaccompanied-migrant-child-workers-exploitation.html?unlocked_article_code=1.NE0.rVW0.ig0MjKFj6LaZ&smid=url-share,https://www.nytimes.com/interactive/2023/12/28/us/migrants-children-data.html?unlocked_article_code=1.Jk0.xUv9.GJ_Yfahr5URj&smid=url-share,https://www.nytimes.com/2023/04/17/us/politics/migrant-child-labor-biden.html?unlocked_article_code=1.NE0.BZTS.KFX8DID9iQBC&smid=url-share,https://www.nytimes.com/2023/03/03/insider/searching-for-the-faces-of-child-migrant-labor.html?unlocked_article_code=1.Nk0.es3f.6KeZXcwNoJF5&smid=url-share,,,,
Body Count: Extrajudicial Killings in Bangladesh,Single project,Netra News,Small,"Nazmul Ahasan, Tasneem Khalil, Peter Aldhous, Soo Oh, Yoli Martinez, and five unnamed journalists and researchers based in Bangladesh.","Nazmul Ahasan is currently a reporter for Bloomberg News, but he tackled the project as a graduate student at UC Berkeley.

Tasneem Khalil is the editor-in-chief of Netra News.

Peter Aldhous is a data, science, and investigative reporter based in San Francisco. 

Soo Oh is currently Investigative Editor at The Markup.

Yoli Martinez is currently a software engineer with The Washington Post.

Peter Aldhous, Soo Oh, and Yoli Martinez contributed to the project as lecturers at UC Berkeley Graduate School of Journalism.

Five other journalists and researchers based in Dhaka, Bangladesh cannot be named due to safety concerns.",Bangladesh,"""Body Count: Extrajudicial Killings in Bangladesh"" is an in-depth investigative report by Netra News, analyzing over a decade of extrajudicial executions in Bangladesh since 2009. It documents 2,597 killings by security forces, revealing patterns of targeted violence, especially during election years. We built a sophisticated database searchable by name, location, and the security agency involved, such as police, army, and containing descriptions for each case. We used interactive maps to demonstrate, for example, disproportionate impact on opposition strongholds and Rohingya refugees who live in the country’s southern border area.","While the “Body Count” story did not lead to immediate, explicit changes in Bangladesh's highly censored political landscape, it complements a broader effort by Netra News and other organizations. Their collective reporting drew U.S. attention to certain security agencies' roles in human rights abuses. Notably, this advocacy influenced the Biden administration’s unprecedented 2021 decision to sanction officials from the Rapid Action Battalion, a move that has been widely credited with reducing such killings in the country.

Immediately after the story was published, Foreign Minister Abdul Momen warned that “journalists should not harm the country’s image just for accolades and awards.” 

Adil Sakhawat, a human rights defender with Human Rights Watch, privately provided the following testimony upon request: “In countries like Bangladesh, the very act of documentation is a giant leap towards ensuring justice. Therefore, the ‘Body Count’ story — a first-of-its-kind story in Bangladesh’s history — will advance the pursuit of accountability for extrajudicial killings in Bangladesh under the country’s authoritarian regime. It has laid the crucial groundwork for rights defenders and investigators to continue documenting the abuses. International actors can rely on the analyses to build the case for why the perpetrating security forces deserve serious censure.”","In recent years, Bangladesh has been termed a variant of autocracy, marked by dwindling press freedom and widespread self-censorship. The report in question was feasible only because its publisher, Netra News, operates from exile, and its reporter, Nazmul Ahasan, was abroad. Although blocked in Bangladesh, the website remains accessible through mirror sites powered by Google that authorities hesitate to block. Netra News’ team faces numerous legal challenges, including sedition and defamation.

The project faced significant resource constraints. Ahasan, a UC Berkeley graduate student, worked voluntarily, leveraging university resources for data processing. The dataset, provided by rights groups, was notably inconsistent and disorganized, requiring extensive effort for cleaning and verifying each report of extrajudicial killing.

“I don’t remember to have ever worked with a messier or more disorganized dataset than this,” Peter Aldhous, a professor at UC Berkeley at that time, said of the dataset.","A group of Bangladeshi human rights defenders tracked individual cases, leading to the Capital Punishment Justice Project compiling them into more than 100 different spreadsheets. These spreadsheets, not uniformly formatted, were refined using OpenRefine and manual efforts, and were combined into a dataframe using the R package called readxl.

For mapping, the tidyverse R library generated coordinates against sub-districts. We then retained coordinates with a high confidence level (98%-100%) and manually adjusted outliers. The data was visually depicted using hex grids on Bangladesh’s map, which we calculated using the R package called SF, with vertical bars indicating casualty counts in each region. We used OpenMapsTiles to visualize the map. 

Additionally, JavaScript libraries such as D3 and jQuery were utilized for creating charts and managing a table with over 2500 collapsible rows.","From the project, journalists can learn about the importance of meticulous data gathering, even from varied and inconsistent sources. The use of advanced data processing and visualization tools, such as OpenRefine, R libraries, and JavaScript, is crucial for presenting complex information in an accessible manner. Additionally, the project -- which continued for more than a year before being published -- exemplifies the need for perseverance in investigative journalism, especially in challenging environments where information is restricted or censored.",N/A,N/A,English,https://interactive.netra.news/extrajudicial-killings-bangladesh/,https://gijn.org/stories/editors-pick-top-10-data-journalism-projects-from-2023/,,,,,,
China: The Superpower of Seafood,Single project,The Outlaw Ocean Project,Small,"Ian Urbina, Joe Galvin, Maya Martin, Susan Ryan, Daniel Murphy, Austin Brush, Jake Conley and the staff of the Outlaw Ocean Project","Ian Urbina is the founder and director of The Outlaw Ocean Project. Previously, Ian spent roughly 17 years as a staff reporter for the New York Times and has received various journalism awards, including a Pulitzer Prize, two George Polk awards, and an Emmy. 

Joe Galvin is the OSINT Editor of the team. Maya Martin is a research editor. Susan Ryan, Daniel Murphy and Austin Brush are Research and Investigations Editors at the Project. And Jake Conley is a research editor at The Outlaw Ocean Project.",United States of America,"China's global fishing fleet is one of the world's greatest maritime enterprises ever. It is also a criminal operation, widely dependent on captive deckhands, illegal fishing, and state-sponsored forced labor from North Korea and China’s repressed Xinjiang region. The Outlaw Ocean Project, in partnership with The New Yorker and two dozen other news organizations around the world, published a series of groundbreaking exposés that prosecuted the case, drawing on four years of reporting conducted aboard Chinese ships on the high seas and inside Chinese seafood plants.","After our investigation, EU lawmakers passed two laws concerning [illegal fishing](https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/impact/#eu-lawmakers-passed-china-transparency-resolution) and [import controls](https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/impact/#samira-rafaela-comments-on-eu-legislation) on products tied to forced labor. The CEO of a major seafood company abruptly [resigned](https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/impact/#high-liner-foods-ceo-resigned) after being presented with findings of illegal practices tied to his company’s plants, and several major seafood companies severed ties to the plants and ships we had revealed were using forced labor. A bipartisan commission created by the White House and Congress held a hearing based on the investigation and demanded documents and answers from federal agencies such as NOAA, Customs and Border Protection, and the Department of Homeland Security—and from companies such as Costco and Albertsons. Nonprofits filed legal petitions to the U.S. and Canadian government seeking a ban on imports from companies cited in our reporting. 

Our reporting triggered extensive coverage within the industry press and spurred debate on op-ed pages globally, including in the _[Guardian](https://www.theguardian.com/commentisfree/2023/oct/13/fish-caught-processed-uyghur-forced-labor)_, _[USA Today](https://www.usatoday.com/story/opinion/2023/10/11/us-seafood-china-human-trafficking-uyghur-forced-labor/71127786007/)_, _[Chicago Tribune](https://cdn.theoutlawocean.com/investigations/china/pdf/coverage/chicago-tribune--op-ed-is-the-seafood-you-buy-connected-to-worker-abuses.pdf)_, and _[Globe and Mail](https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/reach/#the-globe-and-mail)_. Our work got vastly more traction than such reporting typically does, because of the collaborative way it was [disseminated](https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/reach/): in 12 languages, 30 countries, and more than 140 news outlets. Articles based on our reporting attracted a combined readership of over 200 million.","Our work stands out for its scope, complexity, and reach. For four years, we investigated more than 700 Chinese ships, creating a public database that traced their entire supply chain from catch and crime to consumer. No western journalists had ever reported from onboard Chinese ships or traced how seafood caught and processed with forced labor moves through the supply chain to consumers. No journalists, advocacy groups, or academic researchers had ever documented that Xinjiang and North Korean forced labor were being used so extensively in the seafood exported to most of the world’s biggest grocery chains and restaurants—and to the U.S. federal agencies that supply public schools, military bases, and food pantries. 

When we launched this investigation, we focused on China for two reasons: first, because it has become the undisputed superpower of seafood, with more ships and more processing capacity than any other country; and second, because its open-water fishing fleet engages in more abuse and criminality than any other country’s. These factors made China an unusually tricky journalistic target but a uniquely important one. Our investigation tied 357 of the 751 ships we studied to human-rights or environmental violations.","We used a full arsenal of tools, techniques, and technologies to do our reporting. We used satellite data to locate Chinese fishing vessels on the high seas, we spent extensive time on the water searching for them, and when we found them we deployed remote submarines and infrared drones to film them. When approaching them, we initially communicated by radio, often using translators. When captains got spooked and fled the scene, we trailed them in fast skiffs until we got close enough to throw aboard plastic bottles containing pens, cigarettes, hard candy, and interview questions, which allowed deckhands to quickly read our messages and understand our mission. In turn, they would write replies and toss the bottles back to us.

For our land-based reporting on China’s seafood-processing industry, we culled information from cell-phone footage, state media reports, and company newsletters. Additionally, we visited grocery stores around the world to extract codes from seafood packages, which we tied with export reports to trace them to processing plants in China. We also used trade data to tailor individual stories for a dozen countries and partnered with major outlets in those countries to produce stories that would speak to their audiences.","This investigation offers many lessons for reporters. They include: How to connect the dots between crimes at sea and consumers on land. How art can help diversify and amplify journalism. How it’s worth trying to collaborate with as many news outlets as possible, even if those collaborations take time to develop. How to use open-source content, such as cell-phone footage posted to social media, to get eyes into places that are otherwise difficult to access. How to afford yourself greater credibility when presenting your findings, by making full-text versions of all of your interviews available, and by publishing explanations of your methodology. How to avoid demoralizing readers with your reporting by offering them empowering guidance for action suggested by experts and advocates. And, finally, how—by creating an “Impact” page—to enable other reporters, lawmakers, and law enforcement to track the outcomes of an investigation and even advance it.",N/A,N/A,"English, French, Korean, German, Uyghur, Dutch, Finnish, Italian, Spanish, Portuguese, Croatian, Vietnamese, Turkish, Greek, Chinese",https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/a-fleet-prone-to-captive-labor-and-plunder/,https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/the-uyghurs-forced-to-process-the-worlds-fish/,https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/bait-to-plate/,https://www.theoutlawocean.com/investigations/china-the-superpower-of-seafood/discussion/,,,,
Community mapping - tracking the health of our high streets,Single project,"The Courier, Press and Journal",Big,"Lesley-Anne Kelly, Emma Morrice, Ema Sabljak","The DC Thomson data team blend traditional journalism with cutting edge methods to create engaging data journalism.

We strive to democratise data for our readers and make it more accessible using data visualisation, graphics and scrollytelling to explain data in creative ways.",United Kingdom,"With the effects of the pandemic, the cost of living crisis and inflation hitting us all, we wanted to quantify the direct impact to the Scottish high streets and vacancy rates of retail spaces.

The only problem is that no data existed.

We therefore got out on foot in four cities and completed a census and then created hand drawn maps of every retail unit as well as interactive floor plans of every shopping centre.

We then analyse the data and publish regular reports on the health of our high streets.

And now we use the date to drive change.","Since launching we've heard from multiple sections of our community:

* our tracker articles all include a callout for crowdsourcing and we've had great engagement from readers and local business owners wanting to be involved in the project and keep the data up to date.
* we've heard from local government officials and council leaders within our patch telling us that their planning and urban development departments had been using our data.
* we've heard from local government officials outside our circulation area looking for support and advice on setting up similar projects in their area.
* we've heard from charities and community groups telling us they've been using our data.
* we've heard from school pupils asking to use our data in geography classes and looking for support and advice to complete similar projects on a smaller scale.
* we've heard from academics asking to use our data in Government reports.
* we've heard from university staff looking to use our data for teaching purposes.

We have a summary report available internally with up to the minute data for reporters to provide context for the regular articles they produce regarding store openings and closures and the comings and goings of the high streets across our patch.","This data does not exist elsewhere. There may be small snapshots of data on vacancy rates, and there may be some held privately, but there is no freely available up to the minute data. The number of people who have contacted us since releasing it, tells us that this was a problem.

Whilst we may technically qualify as a large newsroom, the data team is three individuals and resources are very limited.

This project is very resource intensive for such a small team - for each of the four cities we completed a census of the 1,809 retail units across 41 high streets and 8 shopping centres. After the initial census we complete a census once per quarter to quality assure the data and receive regular updates from reporters when they report on store openings and closures, in order to keep the tracker articles as up to the minute as possible.

For the vacant units across our patch we collated data on how long they have been vacant, which involved trawling through our archives, Companies House records, various social media accounts, speaking to estate agents etc in order to compile the best available data.","The base map data of units were hand drawn in Geojson.io. The Geojson was then imported to google sheets where the data for each unit was collated, and then Flourish pulls through the data from google sheets to create the interactive maps.

Interactive floor plans were drawn and edited in Figma. The associated data for the shopping centre units was collated to the same google sheet as above and SVGs of the floor plans were pulled through to Flourish and joined to the data from google sheets.

Google sheets was used to analyse all of the data and create aggregate data. Data from here was drawn through to Flourish to create all associated charts.

For the four links I have provided the Dundee project but links to all of the articles (six tracker articles, four launch articles, and two follow up articles) can be found here:

Dundee high street tracker: https://www.thecourier.co.uk/fp/news/dundee/4488541/dundee-shops-track-empty-occupied-units/
Dundee shopping centre tracker: https://www.thecourier.co.uk/fp/news/dundee/4489747/dundee-shopping-centres-track-empty-occupied-units/
Dundee launch article:
https://www.thecourier.co.uk/fp/courier-investigations/4455304/shopping-dundee-high-street-investigation/
Dundee quarterly update:https://www.thecourier.co.uk/fp/courier-investigations/4817412/dundee-shops-vacancy-rate-november-2023/

Aberdeen high street tracker: https://www.pressandjournal.co.uk/fp/news/aberdeen-aberdeenshire/5834641/aberdeen-shops-track-empty-occupied-units/
Aberdeen shopping centre tracker: https://www.pressandjournal.co.uk/fp/news/aberdeen-aberdeenshire/5834686/aberdeen-shopping-centres-track-empty-occupied-units/
Aberdeen launch article: https://www.pressandjournal.co.uk/fp/pj-investigations/5852143/aberdeen-shopping-high-street-investigation/
Aberdeen quarterly update: https://www.pressandjournal.co.uk/fp/news/6227716/aberdeen-city-centre-quarterly-update/

Perth tracker: https://www.thecourier.co.uk/fp/news/perth-kinross/4815732/perth-high-street-shops-track-empty-occupied-units/
Perth launch article: https://www.thecourier.co.uk/fp/news/perth-kinross/4829091/shopping-in-perth-how-healthy-are-our-high-streets/

Inverness tracker: https://www.pressandjournal.co.uk/fp/news/highlands-islands/6280213/inverness-city-centre-thriving-or-in-trouble/
Inverness launch article: https://www.pressandjournal.co.uk/fp/news/highlands-islands/6280213/inverness-city-centre-thriving-or-in-trouble/","If data doesn't exist - create it. If you have spotted the lack of available data, chances are other people have too and will see the value and this will drive engagement with the project.

This project gets to the heart of **local** data journalism - unless massive resources were available it would be very difficult to replicate this at a national scale, but as we have a smaller circulation area we can manage the level of resource involved in maintenance as we live and work in the areas we're tracking, providing something of significant value to our local communities.

The collation of this data also has benefits to our newsroom - reporters can access a summarised version of the current analysis to use in any stories they publish about the high street which adds context for our readers. Reporters have also used the data for story leads (for example a story on why a unit had been vacant for 13+ years). This all adds exclusive benefits that can only be sourced from our data,

Also, never underestimate the benefits as data journalists of getting out from under the spreadsheets and getting out into the world to collect physical data.",N/A,N/A,English,https://www.thecourier.co.uk/fp/courier-investigations/4455304/shopping-dundee-high-street-investigation/,https://www.thecourier.co.uk/fp/news/dundee/4488541/dundee-shops-track-empty-occupied-units/,https://www.thecourier.co.uk/fp/news/dundee/4489747/dundee-shopping-centres-track-empty-occupied-units/,https://www.thecourier.co.uk/fp/courier-investigations/4817412/dundee-shops-vacancy-rate-november-2023/,,,,
Con Copia Oculta,Single project,"Data Cívica, Intersecta",Big,"Mónica Meltis, Estefanía Vela, Alicia Franco, Itzel Soto, Monserrat Perez, Carolina Torreblanca, Laura Atuesta, Itxaro Arteta, Alejandra Nava, Jorge Andrade, Pablo Martínez, Basem Hamza, Cassandra García, Luis Chávez, Pamela Benítez, Verónica Petraca, Gabriela García, Pilar Eunice Medina Rosales, Paris Martínez","**Data Cívica** es una organización feminista que utiliza los datos y la tecnología como herramientas para la defensa de Derechos Humanos. Buscamos impulsar el cambio social, promover y defender los derechos humanos, reducir la brecha en el uso de datos y tecnologías, además de fortalecer a la sociedad civil y al periodismo del país y la región.
**Intersecta** es una organización feminista que impulsa la igualdad en el ejercicio efectivo de los derechos humanos a través de la investigación y la promoción de soluciones de política pública redistributivas, interseccionales, no punitivas y basadas en evidencia al problema de la discriminación.",Mexico,"Entre los mails del Ejército filtrados por Guacamaya, encontramos una base de datos de asesinatos, desapariciones o ataques armados que atribuyen al crimen organizado. Tras analizarla, descubrimos que el gobierno actual considera que 59% de personas asesinadas en México son resultado de la guerra contra el narcotráfico y las clasifica como delincuentes, aunque no tenga información sobre su identidad, si llevaban armas o drogas. Publicamos un informe con gráficas, un reportaje, un scrollytelling de un caso de abuso de poder, y la base con más de 50 mil hechos violentos, después de un trabajo manual de borrarle los datos personales.","Nuestro reportaje fue reproducido íntegramente en 8 medios nacionales y locales, y colaboramos con 9 medios locales para el análisis de la base en sus estados. En total hubo casi 200 menciones en medios impresos y audiovisuales de México y extranjeros, dimos entrevistas y 13 columnistas escribieron sobre nuestra investigación. También otros reporteros usaron “la base oculta” que publicamos para sus propios análisis y notas. Lo cual demuestra que tuvo un gran impacto en la opinión pública.
Desafortunadamente, el presidente descalifica constantemente el trabajo periodístico y cualquier noticia adversa a su gobierno dice que es mentira, además de que no está dispuesto a modificar su estrategia de seguridad usando a los militares.","Las autoridades de seguridad mexicanas han negado que tengan ese registro cuando se les pregunta públicamente o incluso a través de la ley de transparencia, por eso la llamamos “la base oculta”. La guerra contra el narcotráfico utilizando al Ejército empezó hace 17 años, ya abarca 3 presidentes de distintos partidos políticos, y el último dice ser todo lo contrario al primero, pero esta investigación probó que su estrategia de seguridad es la misma, también es igual su manera de registrar la violencia y criminalizar a las víctimas de homicidio sin investigaciones ni pruebas. Es más: ha militarizado gran parte del país y el Ejército controla aeropuertos, aduanas, construcciones, obras turísticas, etcétera. México no sólo tiene una crisis de violencia, sino también de violaciones de derechos humanos por parte de autoridades. El presidente actual, además, califica como “opositores” a los periodistas y dijo que la información filtrada por los hacktivistas de Guacamayas era mentira.","Desarrollo SQL para descargar múltiples bases de datos de los correos electrónicos filtrados por Guacamayas. Análisis de datos en R para encontrar patrones en la calidad de los registros de hechos violentos, con lo que detectamos que tienen menos datos cuando hay alguna autoridad involucrada, lo que muestra el interés de ocultar información. También R para crear visualizaciones y automatización de análisis para generar reportes particulares para distintos estados. Desarrollo web de un sitio propio y animación en forma de scrollytelling de un caso de abuso de autoridad que quedó en los registros oficiales como si personas inocentes hubieran sido delincuente. Software de procesamiento de lenguaje y machine learning para borrar datos personales de la base de datos con más de 50 mil descripciones de hechos.","Que más allá de las revelaciones inmediatas y escándalos descubiertos por el hackeo masivo de Guacamayas, con tecnología y análisis de datos se pueden entender las entrañas de gobierno y la narrativa que están utilizando para justificar la guerra contra el narcotráfico y la muerte de miles de personas que no investigan si son culpables de algún delito o inocentes. También que se puede presentar la información en diferentes formatos y con distinta profundidad de análisis para que sea atractiva, interactiva, y llegue a distintos públicos. Además la importancia de la responsabilidad que tenemos en no divulgar datos personales que pudieran implicar un riesgo para las personas, pero encontrar la manera de hacer disponible una gran cantidad de información que servirá a otras personas en el periodismo o la academia.",N/A,N/A,español,https://concopiaoculta.org/la-bitacora-de-la-guerra,https://docs.google.com/spreadsheets/d/1Jeg0ZrJJzTRhE8HJf6j1e0OyeOXwi0Bth2zlsxRHOxA/edit#gid=0,https://www.milenio.com/videos/milenio-tv/bote-pronto/data-civica-presento-proyecto-investigacion-guacamaya-leks-bote,,,,,
"Despite lawsuit, Casino Group still sells beef from an Amazonian Indigenous land",Single project,[InfoAmazonia](https://infoamazonia.org/sobre),Small,"Fábio Bispo, Juliana Mori, Thays Lavor","**Fábio Bispo**: Investigative reporter for InfoAmazonia. He focuses on political coverage, public transparency, data journalism and environmental issues. In more than a decade of work, he has collaborated as a freelancer for various publications, such as Estadão, Folha de S. Paulo, Revista Piauí, Intercept Brasil, Congresso em Foco, Agência Pública, among others.

**Juliana Mori**: Juliana Mori is a journalist specializing in audiovisual productions and geospatial data visualization. Co-founder and editorial director of InfoAmazonia.

**Thays Lavor**: Data project coordinator.",Brazil,"The report found that the French Casino Guichard-Perrachon continued purchasing meat from cattle raised on Brazilian indigenous lands for their stores, despite a complaint under France's so-called Vigilance Law. This law mandates large companies to prevent serious human rights, health, safety, and environmental violations by their subsidiaries and contractors. Analysis of 500,000 cattle movements from two JBS slaughterhouses supplying Casino supermarkets revealed meat from illegal farms within the Uru-Eu-Wau-Wau Indigenous Land. JBS admitted to the illegalities and suspended these suppliers. Cattle ranching in this region has led to violence and risks the extinction of isolated indigenous groups.","This report was republished by more than 20 organizations in Portuguese and English and prompted debates about the invasion of indigenous territory by cattle ranches. The report's evidence and JBS's confession that animals raised in the protected territory ended up in the Casino Group's stores discount the arguments of the French group, which has informed the French courts that it maintains strict monitoring of its suppliers. 

The report was also cited by different organizations, such as Mighty Earth and Rainforest-Action-Network, in complaints calling for the cancellation of JBS's IPO on the NYSE, pointing out that the meatpacker's activities are related to human rights violations and deforestation in the Brazilian Amazon.

Following the publication of the report, the Coordination of Indigenous Organizations of the Brazilian Amazon (COIAB) said it will take the evidence to the French courts to prove that the Casino group violated its country's laws by buying JBS cattle from illegal ranches,","The Uru-Eu-Wau-Wau Indigenous Land has been one of the hardest hit by the huge global demand for meat. The invasions of the Indigenous Land to supply large meatpacking plants, as we have shown, have the strong support of local politicians who have been pressuring the indigenous people for years. In 2020, indigenous leader Ari Uru-Eu-Wau-Wau was murdered. Although the indigenous people linked the case to the invading farms, the Federal Police ruled out a link between the crime and cattle ranching. The indigenous people continue to be threatened and surrounded by invaders with the support of local politicians and businessmen.

The evidence for this report was only possible after we managed to structure the data on animal movements, which is hidden by both the companies and the Brazilian government [which owns 25% of JBS shares](https://infoamazonia.org/2023/05/24/principal-acionista-da-jbs-bndes-ignora-desmatamento-na-producao-de-carne-enquanto-pune-produtores-menores/).

The data work required a lot of technique to sort out who bought from whom until we were able to prove that the ranches inside the indigenous land sold to direct suppliers. JBS only confirmed the illegalities after we showed them all the evidence.

During the investigation of this report, JBS tried its best to hinder our work, questioning the veracity of the information.","The key information for this investigation was the data from the [Rural Environmental Registry (CAR)](https://www.car.gov.br/publico/imoveis/index) provided by the producers at the time the animals were transported. With the CAR data marked on the animal transportation documents (GTAs), I was able to georeference each movement and find out from which farm the animals left. 

For this work [we used the Python and applied the Pandas library](https://colab.research.google.com/drive/1ly85k2kcJk-QNAonTwNQbEsXXgAnyjgw#scrollTo=feo-luVJjxbM) to organize the data and [reassemble the supply chain](https://lh5.googleusercontent.com/REkvFfmLM5v9j2qP1cUuLp_yc21UGRU-haF1oShOjC52xo10pIw3YrErenhof3TEkQBMLQrFGSO52i_9I7TipUSbebf9x3fVfIG8Igm7Fsm52jKeeHQgUwP80Tlh3Hj2qMG7X6j45dIkhnmbpaALR5A) of the farms, linking who JBS bought from and who JBS's supplier bought from, and so on.

After structuring the data, we put together a PowerBI, which is also available on our [transparency](https://infoamazonia.org/2023/06/29/como-investigamos-a-cadeia-de-fornecimento-do-grupo-casino/) page (in Portuguese), so that readers could see how the information was structured. This database brings together 500,000 cattle movements and is a rich source of new research, since we only focused on ranches on indigenous land. It is possible to identify ranches in other protected areas on this database (which could generate new reports).

For the geospatial analysis [we used the QGis tool](https://lh3.googleusercontent.com/noKk2QuNjdAnRtKHdXADNZMv8RklJeRG-xnDQk1KXa9tzmw_Xm02Xuu1kEsdoHzhNUIXQKRS059ihUublcTvN1ZvqZaOsorloiKiz4SIlBKmeuDouljqxkR80Q70mpeYxhhsJB7Kasbkxar4UBIqE3g), with the data from the CARs identified and overlaid on the indigenous land, and cross-referenced with information on deforestation and environmental embargoes, to broaden the context of each property.","Este trabalho mostra como o jornalismo investgativo guiado por dados pode se valer de diferentes técnicas para a construção de uma narrativa. A ideia de remontar a cadeia de abastecimento dos frigoríficos parecia algo impossível, e que acabou se revelando viavel com uso da linguagem de programação. Mas mesmo tendo os dados das guias de transprte animal (GTAs) ainda assim não teríamos certeza de onde os animais saíram sem fazer a análise geoespacial. Por fim, a confirmação da apuração com os personagens envolvidos, a exemplo da nossa insistência com a JBS, mostra que ouvir o outro lado não deve nunca ser algo meramente protocolar.",N/A,N/A,English,https://infoamazonia.org/en/2023/06/29/despite-lawsuit-casino-group-still-sells-beef-from-an-amazonian-indigenous-land/,https://www.mightyearth.org/wp-content/uploads/230707_Press-Statement_Casino-Case.pdf,https://docs.google.com/spreadsheets/d/1DIfpq8BnAeXLJKJWLjjNQxpLnfVypd__dLGC0OOkPT0/edit?usp=sharing,https://forestsandfinance.org/wp-content/uploads/2023/08/RAN-2023-SEC-Complaint-Submission-re-JBS-Rainforest-Action-Network-Aug17-2023.pdf,,,,
El trabajo en riesgo ante un calor que no es normal (Work At Risk in the Face of Unseasonable Heat),Single project,"El Surtidor (Paraguay), El País (Spain)",Big,"Editorial direction and reporting: Jazmín Acuña
Reporting: Romina Cáceres, Juliana Quintana, Josué Congo and Maximiliano Manzoni
Visual edition: Jazmín Troche
Product Editor: Alejandro Valdez Sanabria
Coordination with El País: Lorena Arroyo
Documentation: Daniela Mohor and Joan Royo (El País)
Visual design and illustrations: William Matsumoto, Naoko Okamoto and Lorena Barrios
Data assistant: Sara Campos
Photography: Elisa Marsal and Sandino Flecha
Video: Milena Coral and Diego Quintana","El Surtidor is an independent digital outlet that covers the climate crisis, disinformation and gender inequality. The team has earned recognition for its highly visual work and how their journalistic content adapts to social media. The newsroom is made up of reporters, designers and illustrators whose aim is to serve young audiences. Jazmín Acuña is editor in chief and co-founder. Alejandro Valdez is co-founder and with Jazmín Troche leads the product and visual work of the outlet. Romina Cáceres is deputy editor. Juliana Quintana and Josué Congo focus on gender inequities and disinformation. Maximiliano Manzoni follows climate change related issues.",Paraguay,"An experiment with delivery drivers in Paraguay’s capital, one of the hottest areas in Latin America, reveals health and safety problems due to rising temperatures aggravated by the climate crisis. The research makes an innovative use of core body temperature monitors to measure heat stress in these urban workers. Reporters combined the findings with weather data to evaluate health risk due to heat. A leading scientist in heat and work studies helped interpret the data. The reporters also gathered scientific literature, public information and carried out interviews to show how high temperatures affect workers’ health and their work performance.","Our project accomplished impact on three levels. First, it engaged the main sources of research in meaningful ways. Five delivery drivers voluntarily agreed to put on core body temperature monitors during their work hours. They did so because they wanted to back up with data their demands to have access to drinking water and resting spots in the city during heat waves. It is worth mentioning the collaboration with one of the leading scientists in heat and work studies. Andreas Flouris became interested in our research and willingly suggested ways of improving our methodology. 

Second, we managed to convene a multi-stakeholder space to present the findings of our work. Delivery workers, representatives from the Ministry of Labor and the Ministry of Health, a senator, two deputies, a doctor and two urban designers gathered in our newsroom for a two-hour conversation where they discussed preventive and mitigation measures. 

Third, the Labor Ministry issued safety measures for workplaces during heat waves just a few weeks after the release of our findings. It is the first such warning we are aware of from a government entity. The warning was released by the office of the director who attended our multi-stakeholder space.","Although Paraguay is one of the most vulnerable countries in South America to the impacts of climate change, El Surtidor is the only news outlet that has had specialized in-depth coverage of the issue for at least four years. In this context, extreme weather events like heat waves may not seem unusual in the public agenda. But we put all of our efforts into unveiling how they are not normal. We do so by choosing angles to our stories that manage to bring the issues of the climate crisis closer to people and show them how its effects can affect their daily lives. 

It is worth mentioning the relative accessibility of the technology we used to measure the impact of heat on health and work performance. While newsrooms are scrambling to see how to incorporate AI technologies into their work, our focus remains on putting the issues that are relevant to our audience at the forefront of our decision-making processes. We also want to be strategic and choose technologies that are within our reach and are useful to answer the questions of our research.","Our experiment had two phases. In the first phase, we measured delivery workers’ core body temperatures by putting monitors on their chests. We wanted to see if they suffer heat stress during their work hours on hot days. The monitors are usually used by athletes, but they have become more common in some industries like construction work to prevent heat injuries. 

In the second phase, we measured the wet bulb globe temperature (WBGT). According to scientists, it is the best way to measure health risk due to heat. We did so by requesting historical weather data (temperature, humidity, wind speed and radiation) to the Weather Services on the same days the delivery workers carried the monitors. We used a tool by the Heat-Health Organization to calculate the WBGT. 

With the data from the monitors and the WBGT we confirmed that the urban workers are exposed to highly and extremely risky heat while working and that their bodies are in danger of short and long term injuries associated with high temperatures.","Our research is an example of how the coverage of the climate crisis can connect to people’s daily lives with data. It is key that newsrooms find ways to narrate probably the most important issue of our time to the public, and they can do so by looking at the aspects that people care about like work and general wellbeing and how these will be affected by climate change. 

The study also shows how to move beyond reporting on the problems and help contribute to thinking about the possible solutions. The work not only summarizes the effects of climate change on health and labor. It also proposes ways of addressing those impacts. 

Our work serves as a reference of how journalism and academia can collaborate to expose the dangers of climate change in a rigorous, innovative manner. A lot can be done with accessible, cheaper technology to measure data that otherwise would not be available. Finally, journalists can learn from our effort to spark dialogue around the findings of the work with key stakeholders.",N/A,N/A,Spanish,https://elsurti.com/futuros/especial/el-trabajo-en-riesgo-ante-un-calor-que-no-es-normal/,https://pulitzercenter.org/stories/work-risk-face-unseasonable-heat,https://elpais.com/america-futura/2023-11-21/ser-repartidor-un-trabajo-de-riesgo-en-un-planeta-cada-vez-mas-caliente.html,https://elsurti.com/fotorreportaje/2023/11/20/como-es-trabajar-en-una-de-las-ciudades-mas-calientes-de-sudamerica/,,,,
Faça o seu ranking,Single project,PÚBLICO,Big,Rui Barros,"Rui Barros: jornalista de dados licenciado em Ciências da Comunicação. Começou na Rádio Renascença, onde se lançou no jornalismo de dados. Em 2020 integrou a equipa do jornal Público, onde continua à procura de histórias em bases de dados e a criar news applications. É professor de Jornalismo de Dados e Web Development.",Portugal,"Há mais de 20 anos publicamos aquilo que ficou conhecido como o “Ranking das Escolas”. Aquilo que começou por ser uma ordenação das escolas com base na sua média evoluiu com os anos, aumentando a sua complexidade e criando outros rankings dentro do ranking para parâmetros como superação face ao contexto social. Também por isso, todos os anos, este suplemento é criticado por algo que é muito familiar a jornalistas de dados: analisar dados implica fazer escolhas. Por isso este ano decidimos dar essa possibilidade aos nossos leitores: se fosses tu a desenhar o ranking das escolas, que critérios escolherias?","O trabalho teve um enorme impacto porque este é um tema “quente” todos os anos, com muitas críticas aos jornais por divulgarem estes dados - que, tendencialmente, mostram os colégios privados em primeiro lugar. Mas aqui tínhamos um algoritmo que permitia articular os pesos que os leitores achavam que cada um dos seis grandes indicadores das escolas merecia: classificações nos exames nacionais, classificações internas, a capacidade de superação face ao contexto socioeconómico, a taxa de conclusão no tempo esperado quando comparada com alunos do resto do país com perfil semelhante e a oferta de planos e actividades extra-curriculares de cada escola, assim como a respectiva localização geográfica.
Esta possibilidade oferecida teve um impacto enorme, especialmente porque mostrou que há milhares de “rankings” possíveis, mas, acima de tudo, de que é ok que assim seja, porque cada pai/encarregado de educação valoriza coisas diferentes.","O trabalho foi metodologicamente complexo de realizar porque é sempre difícil encontrar uma forma estatisticamente correcta de misturar dados que não estão dentro da mesma unidade. Optamos por recorrer, por isso, ao número de desvios padrões face à média que determinada escola estava em cada um dos indicadores. Mas isso obrigou a um longo script e de uma análise detalhada de cada um dos indicadores, bem como uma discussão com inúmeros especialistas na área da educação e na estatística para chegar a um algoritmo que considerássemos correcto e justo.
Outro desafio foi encontrar uma forma de tornar um trabalho complexo, que na prática permitia “jogar” com 41 variáveis e subvariáveis, em algo simples e de fácil utilização. Como uma grande parte dos cálculos estavam a acontecer no frontend, foi também um enorme desafio fazer com que a página fosse leve e veloz como desejávamos.
O trabalho com a directora de arte do jornal foi fundamental, principalmente quando encontramos os desafios impostos pelo mobile.","A análise de dados recorreu à linguagem de programação R, bem como ao Excel, que é sempre o ponto de partida do Ranking das Escolas pela necessidade de vários jornalistas trabalharem em cima dos dados. Tudo o resto foi feito com recurso a Svelte. Quisemos tirar partido da sua reatividade porque o que tornava o exercício interessante era precisamente perceber o que acontecia ao ranking a partir do momento em que alteramos um indicador. Mas essa reatividade acabou por ser também um desafio que fazia com que os cálculos fossem executados no frontend diversas vezes. Tivemos que encontrar algumas estratégias para evitar problemas relacionados com isso.","Como jornalistas tendemos muitas vezes a menorizar a capacidade da nossa audiência, nomeadamente quando o tema envolve números e estatística. E sim, a literacia estatística, numérica e visual da nossa audiência pode ser um forte aliado. Mas por vezes a audiência não precisa desses conhecimentos para entender que se valorizar diferentes coisas, aquilo que é uma lista ordenada - e nós, jornalistas, tendemos a adorar uma lista ordenada porque esta dá-nos um título e um lead - pode mudar. Ou, se quisermos complexificar a linguagem, que dar diferentes pesos a diferentes coisas num algoritmo dá-nos resultados muito diferentes. Mas acima de tudo acreditamos que criamos um instrumento muito útil para pais e educadores.",N/A,N/A,Português,https://www.publico.pt/interactivos/faca-seu-ranking/,,,,,,,
"Ferrogrão railway line will affect at least six indigenous lands, 17 conservation units and three isolated tribes",Single project,"InfoAmazonia, O Joio e O Trigo",Small,"Directed by: Juliana Mori
Coordination and data review: Thays Lavor 
Editing: João Peres 
Reporting and data analysis: Leandro Melito
Dataviz: Carolina Passos
Social media: Luiza Toledo
Design and illustrations: Júlia Lima and Rachel Gepp","Leandro Melito is journalist with 13 years of experience in reporting and editing content in the Environment, Human Rights, International, Politics and Culture sections. As a freelance reporter have worked with several digital outlets such as InfoAmazonia, O Joio e O Trigo and Brasil de Fato covering social and environmental issues, with a special focus on the Amazon since February 2022. Worked as a news editor for the Brasil de Fato website for two years (2020 - 2022) and as a reporter for Empresa Brasil de Comunicação (EBC) for six years (2012 - 2018)",Brazil,"A route of nearly 1,000 kilometres of railway that will traverse the centre of the Brazil through protected areas and indigenous territories that are also home to isolated tribes. This is the Ferrogrão project (EF-170), a monumental undertaking that is the brainchild of major soybean and maize producers in West-Central Brazil, which promises to bolster the new outflow route through the country’s Arco Norte and reduce costs. Exclusive survey reveals a much greater impact than estimated by agribusinesses. Ministry of Indigenous Peoples shows concern and demands consultation, but Lula’s minister of transport is optimistic about project launched by Bolsonaro.","Published simultaneously on the websites InfoAmazonia and O Joio e O Trigo, the publication reached 4,466 views. In addition to being reproduced by 21 Brazilian websites, with emphasis on Sumaúma, Brasil de Fato and Correio Braziliense, the report was translated by the NGO Madre Brava in a campaign focused on Cargill owners and the impacts of the company's activities worldwide. The project was selected as a finalist in two Brazilian awards: the Cláudio Weber Abramo Award for Data Journalism, in the Investigation category, and the National Transport Confederation (CNT) award in the Transport and Environment category.","The dialogue with the indigenous peoples was fundamental for the development of the analyzes presented in the report and account for the human dimension of these impacts. We worked on the hypothesis of an area of influence of the railway greater than that considered in the original project, based on a technical paper prepared by an indigenous association that represents communities of the Kayapó and Panará peoples. The analysis took as its starting point public data available on the websites of the Ministry of Transport and the Institute National Space Research Institute (Inpe) and has the consultancy of Inpe scientist Luís Maurano to verify the analysis of deforestation data from the Satellite Monitoring Project for Deforestation in the Legal Amazon (Prodes). This data-driven investigation follows the principles of transparency. The original databases and those that resulted from the analyzes carried out by the report, as well as the documentation of the journalistic process, are available on the transparency page published along with the article. This is a social commitment to the reader and also guarantees replicability and use by anyone who wants it. Participate in the process, journalism professionals, data scientists, editors and designers.","This investigation used geoprocessing with QGIS software. To calculate the impacts of the railway, we drew a 50-km buffer along the route published in the Transport Information Database of the Ministry of Transport - determined as the project’s zone of influence - and overlapped it with indigenous lands (ILs) and conservation units to calculate the affected area. Including a 10-km buffer zone around the territories, considered as a protection zone, the railway will have an impact on more than 7, 300 km² of indigenous lands and more than 48,000 km² of conservation units. In the project's zone of influence (50-km buffer) we calculated the deforested area along the planned route based on the deforestation data of the Satellite Deforestation Monitoring Project in the Legal Amazon (Prodes) of the National Institute of Space Research (Inpe), for the period from 2008 to 2022. The report survey found that, in this period of 14 years, more than 95% of the deforested area affecting the territories is concentrated around the municipality of Altamira, and affects the Baú (582.69 km²), Menkragnoti (159.38 km²) and Panará (23.11 km²) ILs, which are home to more than 80% of the indigenous population impacted by the project in the region.","Other journalists can learn from this project to carry out investigations into the impact of large infrastructure works on indigenous populations and conservation units, through the use of geoprocessing tools, such as QGIS, and public databases. The step-by-step investigation and data analysis methods were detailed on the report's transparency page, which allows other professionals to have access to this data and can both learn how the report was made, as well as replicate it in new investigations. The transparency page also lists all the sources that were consulted for the report and explains in what context each of them was approached, which can help other journalists to navigate the topic. This page also points out the limitations of the analysis used, which allows other journalists to make progress on these points in future reports.",N/A,N/A,Portuguese / English,https://infoamazonia.org/2023/07/05/ferrograo-afetara-pelo-menos-seis-terras-indigenas-17-unidades-de-conservacao-e-tres-povos-isolados/,https://infoamazonia.org/en/2023/07/05/ferrograo-railway-line-will-affect-six-indigenous-lands-17-conservation-units-and-three-isolated-tribes/,https://burninglegacy.org/evidence/ferrograo-railway-line-will-affect-at-least-six-indigenous-lands-17-conservation-units-and-three-isolated-tribes/,https://www.youtube.com/watch?v=0Jt2fxG25Cc,,,,
Finding Nigeria’s Forgotten Mass Graves Through Satellite Data,Single project,"1. HumAngle Media
2. Newslines Magazine",Small,"1. Byline: Mansir Muhammed
2. Satellite imagery analysis and visualisation: Mansir Muhammed
3. Field Reporting by: ‘Kunle Adebajo
4. Edited by: Rasha Elass and ‘Kunle Adebajo,","Mansir Muhammed, a Geospatial Intelligence specialist with expertise in geography and environmental management, holds the position of Senior GIS/OSINT Specialist at HumAngle. He specializes in conducting geographic-based investigations related to critical global matters such as conflict, humaniterian crisis, and the environment. Employing advanced scientific techniques and GIS tools, he provides data-driven insights and visual presentations for both media audiences and policymakers. In his role at HumAngle over the past few years, he has conducted open-source investigations and performed data analysis to support news production as the Senior GIS/OSINT Specialist.",Nigeria,"The Mass Graves Project unveils an elusive truth in the decade-long problem of missing persons in northeastern Nigeria, employing satellite imagery and open-source tools to expose secret mass graves concealing victims of extrajudicial killings. Through innovative reporting techniques and data analysis, our investigation not only identified suspected gravesites but also estimated the number of victims in an effort to find the gap in the Nigerian Missing Persons crisis. 

The project employed advanced deductive approach and mathematical computations for investigative journalism. It also documented the process of the investigations.","The report is a part of the documents we presented to the authorities in a series of investigative pieces [https://shorturl.at/tGO13, https://shorturl.at/jnG24, https://shorturl.at/rDQV4]
which we used in hopes of getting justice for victims and their families and those who were still held captive by the government. A few days after our initial efforts, some of the missing people who were unlawfully held in detention were allowed to contact their families. We published a news report about it. Link: https://shorturl.at/rDQV4

Secondly, the approach to investigative journalism through open source data and sateliite investigation was wildely recieved by journalists across the field. Many reaching out to discuss future human rights projects. Global Investigative Journalism Network Africa, did a report on it, featuring it in the headline of their newsletter, and then also published a set of open-source tools encouraging journalists who want to conduct the same investigation.",The report is heavy on innovative technology and data. The type of satellite technology used was originally used for environmental research and this opens up avenues for future investigation for news application. And then the use of a procedural deductive investigative approach to circumvent gaps and limitations in both the availability of data and access to locations of interest.,"Tools include:
* Field data collection 
* Conventional (Google Earth) and Radar satellites (NASA Synthetic Aperture Radar product)
* Geographic Information System (Arcgis: Arcmap and Arcgis pro, Qgis and Google Earth Engine).

Techniques:
* News investigative techniques
* Mathematical computations
* Open Source Intelligence framework
* Geospatial Analysis

The report itself is essentially a methodological approach to using these resources to investigate the issue. It relied on the Geospatial intelligence techniques to gather information which was then interpreted using news reporting style.","They can learn to find meaningful data and resources when conventional methods are limited or unavailable. They can learn to use free and publicly available resources to enhance fact-finding and fact-checks in the course of their investigations. And most importantly, they need to know that these approaches and tools exist and are not beyond them if they give them a chance. With my experience based on the reception I got from the report, many reporters can now see that some of the stories they gave up on, or didn't consider, now seem possible and within reach.",N/A,N/A,English,https://humanglemedia.com/finding-mass-graves/,https://newlinesmag.com/reportage/how-we-uncovered-mass-graves-in-northern-nigeria/,https://humanglemedia.com/into-the-abyss/,https://humanglemedia.com/hello-my-love-giwa-barracks-detainees-finally-allowed-to-make-phone-calls/,,,,
Forest Thieves,Single project,Intercept Brasil,Small,"Fernanda Wenzel, reporter and project creator
Bruno Kelly, photographer
André Uzêda, editor
Júlia Coelho, art
Rodrigo Bento, art
Luiza Drable, video director
Vitã, screenwritter
Marlon Peter, video editor","Fernanda Wenzel is a 14 years experience reporter from Porto Alegre, Brazil, specialised in the investigation of environmental crimes. She is the creator and screenwriter of the documentary BR Acima de Tudo (Brazil Above All) and a former fellow of the Rainforest Investigation Network, where she investigated land grabbing in the Amazon. Her work has been published in outlets like Repórter Brasil, Folha de São Paulo, Valor Econômico, BBC Brasil, Globo Rural, Revista Piauí, Agência Pública, The Intercept Brasil, El País Brasil, ((o))eco, Sumaúma, Diálogo Chino, Mongabay, The Guardian and NBC News. Currently, Fernanda is working in Mongabay.",Brazil,"""Forest Thieves"" is a one year-long investigation published by The Intercept Brasil from November 2023 to April 2024 and which revealed the gears of the land grabbing industry in the Amazon. In this four story series (plus a video-explainer) we showed that it is not a random phenomenon, but a well organized business which involves many actors and a lot of money. We showed who was behind the largest deforestation in the Brazilian Amazon, who are the land grabbers' engineers, how illegal loggers profit over public lands and how Bolsonaro's government gave up protecting 8 million hectares.","After the publication of the serie, in 2023, the Federal Police started a raid against land grabbers in Pará. Among the targets of the investigations were Bruno Heller, mentioned in one of our stories as one of the land grabbers of BR-163 area, and Bianor Dal Magro, the land grabber engineer we revealed in our second story. Even though we can not assure the raids had a connection with our investigation. 

The innovative way in which I combined different databases in this investigation also led me to present my work at the International Journalism Festival, in Perugia, the GIJN 2023's conference, in Gothenburg and the congress of the Brazilian Investigative Journalism Association, in São Paulo. In December 2023, I received an offer to work as a full-time reporter in Mongabay. 

Regarding audience numbers, the stories had 33,731 page views in Intercept's website, 1,797,871 summing Instagram, Facebook and Twitter. The video resuming all the investigation had 9,700 views on Youtube, 65,247 in Instagram and 1,328 pageviews in the website.","This series was produced during Bolsonaro's government, who diminished the repression to environmental crimes and encouraged land grabbers and illegal miners in the Amazon, creating a hostile environment to journalists. Our fieldwork happened a few weeks after the murder of The Guardian's correspondent Dom Philips and of the indigenist Bruno Pereira, which forced us to implement an ever more strict security protocol. We spent most of the trip undercovered, which didn't avoid us being intimidated while taking photos of a sawmill in Rondônia. We had to be even more careful about our local sources, who only could be interviewed off the record and should not be seen in our company. We spent most of the time in small villages in the Deforestation Arch, where the whole economy revolves around land grabbing and illegal mining, and where it was impossible to go unnoticed. Once we were back to the office, we faced legal threats from two of the individuals that would be mentioned in the stories. From the innovation's perspective, we definitely innovated in our third story in which CCCA could estimate the volume of wood illegally taken from the forest in João Bento's glebe.","I used QGIS to analyse geospatial data, specially by crossing the deforested areas with the shapefiles of land registers and environmental fines. That is how I found out who was behind the largest Amazon deforestation, for instance. I also used QGIS to prepare my fieldwork and my security plan, by finding the best routes. During the fieldwork I used geolocation apps like Geotracker and Avenza to guide myself to the specific farms I was investigating and also to track my route and not get lost. I also used the geotracker equipment Stopx to share my location with my editors in real time. I used Open Refine to analyse large data sets like environmental fines, and I read many lawsuits to understand how a particular area was deforested or the history of exploration of a certain area. Social media were also sources of information about the individuals and their role in their communities. Finally, high resolution satellite imagery was crucial both for reconstituting the story of a deforested area (from selective logging to large-scale deforestation and the use of fire) and for estimating how much wood was taken from João Bento's glebe.","They can learn how to tell a story having only a particular area in the map as a starting point. As you add different layers of information on the top of this area, you can get to the individuals and the details that will make that a relevant story. Something that could be treated as just one more deforestation case can become a live illustration of a whole phenomenon like land grabbing, as we did on our first story. I believe the series is also an inspiration to colleagues looking for new ways of covering the same issues. One of them is to dive into the entrails of these dynamics, showing the importance of actors that may pass unnoticed but without whom crimes such as land grabbing couldn't happen. That was the idea behind the second story, in which we showed how ordinary professionals, like a real broker and an engineer, play crucial roles in the land grabbing industry. My work is also an example of how to associate hard data work with traditional reporting techniques. Without the fieldwork and the dozens of interviews I would never get to tell these stories.",N/A,N/A,Portuguese,https://www.intercept.com.br/series/ladroes-de-floresta/,,,,,,,
Ghost Tags: Inside New York City’s Black Market for Temporary License Plates,Single project,"Streetsblog NYC
The New Jersey Monitor co-published parts 1 through 3
Motherboard helped edit and co-published ‘Duped’",Small,"Reporting and writing: Jesse Coburn,
Editing: Gersh Kuntzman","Jesse Coburn is Streetsblog’s investigative reporter. His reporting has received awards from the Sigma Awards, the Hillman Foundation, the Silurians Press Club, and other organizations. Previously he was a reporter at Newsday and an editor at ARCH+. He has also written for the New York Times, the Baltimore Sun, Harper’s, Cabinet and other publications.",United States of America,"This seven-month, data-driven investigation exposed a widespread scheme in which New Yorkers exploited weak regulations in other U.S. states to open sham car dealerships, access government systems for producing temporary license plates, and sell large numbers of those tags illegally in New York City.

Streetsblog NYC, a small nonprofit newsroom, showed how motorists relied on these fraudulent license plates for anonymity while dodging tolls, driving without insurance, and committing more serious crimes. So-called temp tags have been used in shootings, robberies, and hit-and-runs.

Streetsblog further revealed the failure of regulators to stop the illicit trade in these government documents.","The impact of this reporting has been substantial. In New Jersey, legislators unanimously passed a bill to reform the state’s temp tag system, and policymakers unveiled a regulatory overhaul that could shutter dealer compounds identified by Coburn as illegal temp tag mills. Lawmakers in New York City and Georgia also proposed new laws on the issue in the wake of Coburn’s reporting. Auditors shut down dozens of dealerships for temp tag fraud, including some that were highlighted in the series. Other alleged temp tag sellers uncovered by Streetsblog are under criminal investigation.","Dubious paper license plates from far-flung states became ubiquitous in New York City during the pandemic. But, outside of Texas, there had been little substantial reporting of the issue prior to Streetsblog's investigation. No news outlets in New York had figured out who was producing all these paper tags or how they were getting away with it.

To answer these questions, Streetsblog analyzed four years of data on 10,000 car dealerships and combed through 1,000 pages of motor vehicle agency records obtained through public records requests. Streetsblog also interviewed more than fifty people, securing on-the-record accounts from temp tag sellers and buyers who admitted to their crimes.

Streetsblog faced significant resistance to this reporting. Our reporter was barred from a major dealer compound, faced intimidation and legal threats from dubious car sellers, and encountered a wall of silence from industry regulators. Our newsroom has just four reporters and two editors. Overcoming these obstacles and publishing this series is the most significant project we have ever undertaken.","Streetsblog submitted nearly 50 public records requests to obtain data and records on temporary license plates issued by car dealerships in New Jersey and Georgia, two states whose temp tags were especially prevalent in New York City.

Streetsblog used Microsoft Excel to build multiple databases based on that data and those records. The databases contained:
* the temporary license plates issued by every used car dealer since 2019 in New Jersey and Georgia.
* every car dealership in New Jersey and Georgia that had been caught fraudulently issuing temporary license plates since 2020
* the car dealership and temporary tag regulations in five states: New Jersey, Georgia, New York, Pennsylvania, and Connecticut
* the basic temporary tag regulations in 41 U.S. states (nine states did not respond to Streetsblog inquiries)

This data analysis became the foundation of the series by revealing its main findings. First, no-name car dealerships were issuing thousands, even tens of thousands, of temp tags each year while conducting no other discernible business activity. Second, while some of those dealers were getting caught fraudulently issuing tags, the fines they face were small compared to the large sums to be made selling tags illegally.","This project sheds light on the useful public data and records maintained by motor vehicle agencies, which could serve as the basis for many more journalistic projects. It also provides insight into the techniques that reckless motorists use to evade accountability on the road -- a problem that is likely to become only more prevalent as more and more cities around the world adopt automated traffic enforcement tools.",N/A,N/A,English,https://www.streetsblogprojects.org/ghost-tags-index,,,,,,,
Grist_Any Way the Wind Blows,Single project,Grist,Big,Naveena Sadasivam and Clayton Aldern,"Naveena Sadasivam is a senior staff writer at Grist.

Clayton Aldern is a senior data reporter at Grist.",United States of America,"On the edge of Port Arthur, Texas, a majority Black and Hispanic community along the Gulf Coast, sits Oxbow Calcining. Owned by William “Bill” Koch, the facility purifies petcoke, a petroleum byproduct, and in the process emits more than double the sulfur dioxide of an average U.S. coal-fired power plant. Grist reporters got access to a trove of legal documents that laid out years of Oxbow’s internal operations. What they found was even worse than we expected: A corporation committed to exploiting a loophole in the Clean Air Act, and a community left to breathe in the toxic consequences.","For years, residents in Port Arthur have complained about the high levels of air pollution coming from the plant, but with no results. They often wake up to find their cars covered in a thin coat of yellow dust. And the asthma rate in the residential neighborhood surrounding the plant is 70 percent higher than the national average.

Tapping into more than 2.5 years of internal Oxbow operational data, Grist found clear evidence that Oxbow Calcining manipulated its operations, shutting down kilns to lower sulfur dioxide emissions, only when wind blew in the direction of fence line air monitors. The activity created a false profile of Oxbow’s pollution levels, allowing the company to avoid penalty and to pump vast amounts of sulfur dioxide, which can increase the risk of strokes, asthma, and hospitalization, into a neighborhood that is 90 percent Black.

When Grist senior staff writer Naveena Sadasivam and senior data reporter Clay Aldern presented their findings to a former EPA enforcement official, that expert became convinced they had uncovered a clear ""criminal violation of the Clean Air Act."" The EPA has promised to follow up on their findings.","Beyond the rigorous data analysis and reporting, this piece also enabled Grist to pursue comprehensive community engagement. We partnered with the Houston Chronicle and the Beaumont Enterprise, the closest paper to Port Arthur, cutting down our 6,000-word investigation into a 1,500-word version that appeared on the front page of their Sunday papers. 
We also designed and sent postcards to every household and business in the same zip code as the plant, alerting them to the reporting and giving them an avenue to get in touch with any questions or story tips. We were able to place the feature in a 9-minute segment on NPR’s Here and Now. Finally, we hosted a virtual town hall event in late May, bringing together residents, activists, and legal advocates to discuss the community’s path forward in light of Grist’s findings.","Our data analysis relied on underutilized public records from previous legal proceedings. To build a statistical model of Oxbow’s kiln operations, we filed FOIA requests, extracted and cleaned data from the resulting PDFs, and scraped meteorological information from a nearby weather station. Our final dataset included wind speed, wind direction, sulfur dioxide readings from the closest air monitor, and time-locked organizational behavior (i.e. indications of which kilns were on and off at the petcoke plant) at a 5-minute resolution. Using the R programming language and a variety of statistical techniques — including temporal downsampling, random forests, and cross-validation — we were able to build a model of organizational behavior that performed much better than chance.

In doing so, we were able to demonstrate a big effect of southerly winds, higher wind speeds, and the derivative of wind direction on the probability of Oxbow’s kilns being off. Accordingly — and via additional statistical tests and visualizations — we were able to argue that the relationship in question was causal: namely, that operators were modifying the plant’s behavior when the wind was blowing in the direction of the air monitor. We visualized our data using R, Tableau, Adobe Illustrator, and JS/HTML/CSS.","In analyzing large datasets from diverse sources using advanced statistical models, we demonstrated how detailed, data-centric investigations can reveal critical insights into environmental compliance and pollution. Our approach of integrating machine learning techniques with on-the-ground and public-records reporting serves as a blueprint for journalists aiming to investigate local environmental issues with global relevance. Furthermore, our commitment to transparency, by sharing our methodology and data publicly, encourages collaborative and reproducible journalism. This project stands as a testament to the potential of investigative journalism in holding powerful entities accountable and fostering an informed public discourse on environmental matters.",N/A,N/A,English,https://grist.org/project/accountability/koch-oxbow-port-arthur-texas-clean-air-act-pollution/,https://github.com/Grist-Data-Desk/windblow,https://www.wbur.org/hereandnow/2023/03/20/koch-coke-plant-pollution,https://docs.google.com/presentation/d/1fblgjq1t2eoA5AQiwllK5Q2_drwVSrtPyw1ExkfYd0o/edit?usp=sharing,,,,
Ground control,Single project,"Tagesspiegel (Germany), Mediapart (France), Apache (Belgium), BIQdata.pl / Gazeta Wyborcza (Poland), Telex (Hungary), Deník Referendum (Czechia)",Big,"Editorial and research coordination: Gaby Khazalová 

Tech and visual coordination: Hendrik Lehmann 

Data coordination: Helena Wittlich 

Project assistance: Federica Testi 

 

Research in the participating countries: 

Steven Vanden Bussche (Apache, Belgium) 

Alexander Abdelilah (France) 

András Miszur (Telex.hu, Hungary) 

Vilmos Weiler (Telex.hu, Hungary) 

Dominik Uhlig (BIQdata.pl / Gazeta Wyborcza, Poland) 

Michał Iliev (BIQdata.pl / Gazeta Wyborcza, Poland) 

Alicja Gadomska (BIQdata.pl / Gazeta Wyborcza, Poland) 

Danuta Pawłowska (BIQdata.pl / Gazeta Wyborcza, Poland) 

Nina Breher (Tagesspiegel, Germany) 

Corin Baurmann (Tagesspiegel, Germany) 

Hendrik Lehmann (Tagesspiegel, Germany) 

Helena Wittlich (Tagesspiegel, Germany) 

Gaby Khazalová (Deník Referendum, Czechia) 

Data visualization and web development: 

Lennart Troebs (Tagesspiegel, Germany) 

Tamara Flemisch (Tagesspiegel, Germany) 

Kirk Jackson (Tagesspiegel, Germany) 

Katarzyna Korzeniowska (BIQdata.pl / Gazeta Wyborcza, Poland) 

Satellite team: 

Michael Anthony (Vertical 52) 

Marcus Pfiler (Vertical 52) 

Daniel Lanz (Marple) 

Federico Monica (PlaceMarks) 

Michele Luppi (PlaceMarks)","The Urban Journalism Network is a network of European journalists, media outlets and data analysts, that is dedicated to researching common challenges faced by European cities. After winning the European Press Prize 2022, the team behind “Cities for Rent”, an investigation organized within the Arena Housing Network on corporate landlords re-organised into a wider network for investigation of urban problems. The network already published two investigations: about student housing and on the trading of mortgages around the world. In addition, it develops technology for better data-collection and visualization across languages and experiments with AI in order to advance reporting techniques.",Germany,"Affordable housing is becoming scarce in European cities. While politicians and developers suggest solving the housing crisis by building new apartments, there is an underlying problem: not only are rent prices rising, so are land prices. 

Where can affordable housing be built? Who owns unused land? What amount of public land has been lost to selloffs? A team of investigative, local, and data journalists examines these 'ground control' dynamics in European capitals. They're analyzing land price trends, ownership structures, and more, using satellite and aerial images to reveal urban development patterns and increase transparency in urban politics.","The kickoff of our half-year-long project, which started in November, featured 7 articles across five media outlets. By then, we had amassed over 100,000 views across these platforms. The stories were shared several hundred times on the media outlets' social media channels, including Instagram, X, Bluesky, and Reddit, to name just a few. 

Up until the deadline of this award, more than 15 stories on this topic have been published, indicating increasing reader interest and underscoring the importance of housing reportage. Furthermore, all outlets have gained new subscriptions for their publications. 

We are sure that subsequent publications in January and February are expected to have even greater impact, as they will include our findings from satellite analysis and more investigative work. We have already received several requests from politicians and housing activists for access to our collected data on land sales.","There is a significant lack of transparency regarding land ownership and prices. Some members of the urban journalism network have already worked on increasing transparency in the rental markets of Europe and now face a new challenge in the realm of land ownership. So far, there hasn't been any comparable data on land prices or ownership. 

The responses to our FOI requests vary from city to city. While some administrations provide data on their cities' sell-offs of public land, others, like Berlin, deny access to documents regarding public land sales. The German team is considering legal action. 

The difficulties many project partners have faced indicate that the topic of land ownership is worth further investigation. In off-the-record conversations, experts and politicians agree that public land sales are a key area with potential for corruption. 

Committed to utilizing as many innovative technologies as possible, we are collaborating with experts in computer vision, satellite analysis, and AI to analyze and compare urban developments in cities.","Firstly, all visualizations are created using a custom-made graphic management system, enabling every project partner not only to translate the graphics but also to use custom font colors according to each partner's corporate design. The backend was developed using Django and Postgres, while the frontend was implemented with Svelte. 

With the help of two satellite companies experienced in working with journalists, we leveraged the latest computer vision technology to detect new buildings in aerial images. For our satellite analysis, we used impervious surface data to identify urban changes. 

As an initial approach to analyzing spatial data without the need for coding skills, we often use the open-source tool QGIS.","The project is a great example of how collaborative investigation unravels the lack of comparable data. Our research reveals that even the European Union lacks comparable data on land prices. 

We learned to quickly find alternative methods for research, investigation, and storytelling when official sources are uncooperative. 

Furthermore, knowing how to use technology can lead to achievements beyond initial expectations. Many local stories are hidden in large shapefiles on various open data servers. The ability to read and analyze these files is invaluable. For journalists, QGIS is an excellent starting point, with numerous online tutorials available for those with limited technical knowledge.",N/A,N/A,"English, German, Czech, Polish, French, Hungarian, Flemish",https://urbanjournalism.org/,https://interaktiv.tagesspiegel.de/lab/ausgebaut-berlins-explodierende-bodenpreise-machen-wohnungsbau-unbezahlbar/,https://interaktiv.tagesspiegel.de/lab/wem-gehoert-der-boden-wo-berlin-noch-berlin-gehoert/,,,,,
How China is tearing down Islam,Single project,The Financial Times,Big,"By Peter Andringa, Irene de la Torre Arenas, Max Harlow, Sam Joiner, Joe Leahy, Lucy Rodgers, Eva Xiao, Yuan Yang, and Sun Yu.

Drone footage by Marine Zambrano.

Additional reporting by Steven Bernard, Dan Clark, Oliver Hawkins, Eade Hemingway, Ella Hollowood, Alison Killing, Caroline Nevitt and Eri Sugiura.","The FT’s Visual Investigations team is an interdisciplinary group of journalists harnessing computational tools and traditional reporting to break new stories in visual formats. In collaboration with the Visual & Data team and the newsroom as a whole, we focus on stories that use innovative methods and have the greatest impact.",United Kingdom,"China’s oppression of Ugyhur Muslims has been widely documented in Xinjiang. But the government is increasingly restricting Islam across the country, altering or destroying mosques in a bid to reduce the visible presence of religion in public.

The FT examined thousands of mosques using satellite images, finding nearly three-quarters of those with Arabic features have been altered since 2018. The investigation — the first to document the scale of China’s policy — found renovations and demolitions in nearly every province. On-the-ground reporting corroborated this analysis, with FT journalists speaking to communities across the country who are feeling the pressure of Beijing’s crackdown.","The project was a hit with FT readers (one of our most-read visual investigations of the year) and was widely shared: among Chinese-language communities, by human rights groups, and on BBC Radio 4.

While earlier reports have documented specific cases or regional examples of religious repression in China, this is the first time anyone has comprehensively measured its scale. The project documents how these isolated incidents actually underpin a troubling trend — and the importance of paying attention to the infringement of religious freedoms of far more Chinese Muslims than were previously known. The story is an important piece of analysis and helps frame how researchers, scholars and NGOs should approach the subject of religion in XI's China.","This project involved significant behind-the-scenes technical work to identify locations where we might find mosques, count those with Islamic architecture, and look for changes to the buildings over time. The published story is only the tip of the iceberg of reporting and analysis that went into it.

The story required a painstaking manual review of the architecture of 4,450 buildings. We had to guarantee the accuracy of every mosque we counted, a process which involved hundreds of hours of human labelling aided by multiple custom-built tools.

We also didn’t want individual stories to get lost in the data. The FT sent our reporters to visit mosques, speak with community members, and understand how this policy was felt on the ground. We also sent photographers to capture details of the modifications with drone and handheld cameras, helping readers see the changes more clearly for themselves. As with most journalists in China, our in-country staff pursued this story at great personal risk: from past statements and conduct, the government has made clear that it does not want the media covering this policy.","We built an initial dataset of 4,450 potential mosques across China by scraping search results from Baidu Maps, OpenStreetMap, and Google Maps — creating a more accurate and up-to-date list than the official Chinese census. Next, we built a tool to collect satellite imagery for each of these mosques from Google Earth and Planet Labs, aiming to collect a series of images capturing each building at different points in time.

We explored using a machine learning model to classify the architectural styles and/or destruction of each mosque, but concluded any model would have an unacceptable error rate — even if its accuracy was a best-in-class 95%. Instead, we built a labelling tool with a user interface that allowed FT staff to carefully inspect each site over time, spreading the work across a number of reporters and cross-validating their findings.

In total, we identified 2,312 mosques with Islamic architecture. Of these, 1,714 (74.3 per cent) had Arabic-style features removed between 2018 and 2023. Examples of renovations to mosques small and large were found in every region, and in both rural and urban areas.","The project illustrates the importance of human judgement in computational journalism: even when technical tools like AI models can work faster than people, there are many cases where the accuracy of manual data collection is irreplaceable. Instead of racing to implement (often error-prone) machine learning, newsrooms should aim to build tooling that speeds up human workflows but defers to their judgement. We could not have collected this national dataset by manually looking up each mosque by hand — but the combination of data scraping, image collection, and a well-designed labelling interface made it possible to work at greater scale without tradeoffs in accuracy.

The design of this project also demonstrates how good storytelling works at both macro and micro levels: the story is more impactful because it highlights the voices of individuals and communities affected by the policy it documents. Many good data stories can offer a 10,000 foot view, but the truly great ones help readers both see the big picture and imagine themselves in the shoes of the people involved.",N/A,N/A,English,https://ig.ft.com/china-mosques/,,,,,,,
"How diplomats who traffick, exploit domestic workers get away",Single project,Rappler,Small,"Authors: Pauline Macaraeg, Ana P. Santos, Michelle Abad
Consultant: Gerald John Guillermo","Pauline Macaraeg is digital forensics researcher for Rappler. She writes about the developing digital landscape, as well as the spread and impact of disinformation and harmful online content.

Ana P. Santos is an independent journalist who reports on the intersections of gender and sexuality, sexual and reproductive health and rights, and labor migration. She has written extensively about the gendered experiences of the Filipino diaspora in Europe and the Middle East.

Michelle Abad is a multimedia reporter at Rappler. She covers overseas Filipinos, the rights of women and children, and local governments.",Philippines,"A data-driven, cross-border investigation uncovering how erring diplomats have been able to exploit their migrant domestic workers and have gotten away with it for decades, owing to a privilege called diplomatic immunity.

Because of this privilege, it is difficult for workers to seek legal remedies. Documents and reports obtained by Rappler showed that even when the diplomats were charged and courts ruled in favor of the workers, nongovernmental organizations have noted that enforcing judgments remained difficult.","This report created a database of reports of incidents of abuse of migrant workers done by their diplomat employers. Prior to this, there were only separate news reports, court records, and legal journals. We consolidated these accounts into one database, and further enriched it by doing a digital investigation by looking at other available information online.
The investigative piece was listed by the Global Investigative Journalism Network among the [Best Investigative Stories from Southeast Asia](https://gijn.org/stories/2023-best-investigative-stories-southeast-asia/) in 2023.
The Philippines’ [Center for Media Freedom and Responsibility (CMFR](https://cmfr-phil.org/media-ethics-responsibility/journalism-review/immunity-or-impunity-rappler-spotlights-abuse-of-workers-by-diplomats/) also hailed the story for putting a spotlight on this underreported issue. It also said that “the effective presentation of data…allowed readers to explore and understand the numbers in the database.”","This data-driven piece is part of a 4-part series called [The Diplomatic Immunity and Impunity series](https://www.rappler.com/newsbreak/investigative/diplomatic-immunity-impunity-abuses-migrant-workers-series/). Other parts of the series focused on case studies, interviews, and policy issues.

Journalismfund Europe and the Pulitzer Center supported the reporting for the series.","The team created a database of reports of incidents of abuse of migrant workers done by their diplomat employers.

We started with existing separate news reports, court records, legal journals, and personal interviews.

We enriched this database by doing a digital investigation. We looked for other available information online from open-source data, such as case files from NGOs and other news reports. We did keyword and boolean searches to look for these things online, and then we manually curated the results to ensure accuracy.

The team was able to track incidents recorded from 1988 to 2021.","As CMFR noted, instances of abuse done by diplomats are underreported. This project can help journalists from all over the world to be more informed about how systemic the problem is.

The database can also aid journalists in their reporting on this particular issue, since we’ve seen cases involving repeat offenders. Establishing connections like this may help in demanding accountability from erring diplomats.",N/A,N/A,English,https://www.rappler.com/newsbreak/investigative/how-diplomats-traffick-exploit-domestic-workers-immunity-impunity/,,,,,,,
How Spanish victims of gender-based violence likelihood to be protected changes depending on where the complaint is lodged,Single project,elDiario.es,Big,"Raúl Sánchez, Marta Borraz, Ana Requena Aguilar, Victòria Oliveres","Raúl Sánchez: Spanish data journalist covering stories of inequality, gender, health, taxes and urbanism at elDiario.es. He coordinates elDiario.es data team.

Ana Requena: journalist specielized in gender stories. Now, chief gender editor at elDiario.es.

Marta Borraz: journalist covering gender, LGBTIQ and reproductive rights stories.

Victòria Oliveres: Data journalist member of elDiario.es data team. She is specialized in data gathering, analysis and visualization and focused in education, health, environment and gender stories.",Spain,"Will the judge protect you? This investigative journalism project sheds light on the stark inequalities faced by women seeking protection from gender-based violence in Spain. Our analysis, drawing from data obtained from over 500 courts handling violence against women cases, unveils a concerning reality: the likelihood of a judge granting protection varies dramatically based on the location of the complaint. The staggering disparity, ranging from 24% to 100% in the approval of protection orders, underscores the influence of individual judges' positions, case overload, and the shortage of resources and awareness in the judicial system.","We unveiled the protection order rates for every specialized court in Spain, providing women in Spain unprecedented access to data that exposes systemic inequalities in the Spanish justice system. In fact, making this data available was the main goal of this project. By presenting comprehensive data, including interactive cartograms, graphics and maps, we aimed to equip women with valuable insights into the system inequalities they face when denouncing gender-based violence.

Our findings have not only empowered women to make informed decisions but also initiated a self-reflection among judges because “violence is the same in any territory”, as one of our sources explained to us. For example, this investigation allowed judges from different courts to compare their protection rate with all other specialized gender-based courts of the country, their justice area or even their region, which could lead to change of justice policies based on data. 

The investigation reveals that 17 courts conceded less than 40% of protection orders between 2005 and 2022, an indicator that these judges are protecting less women who denounce in those courts.","The data was obtained using an automated process to extract records from all annual reports, using an R script to get individual data from each court. Obtaining detailed and transparent data on protection order rates for every specialized cour in Spain required navigating through the bureaucratic system and a complex understanding of technical nomenclatures of the Spanish judicial system. For example, we had to interview several sources from the statistical offices of the administration of justice to understand the regulatory and data storage changes during the last two decades.

The analysis leads us to prove that the probability of a woman victim of gender-based violence securing a protection order multiplies based on the court she chooses for denunciation. These findings reveal wide disparities of the protection order rate across jurisdictions.

In-depth interviews were conducted with senior judges with years of expertise and judicial experts to understand the reasons behind this gap between different courts among the country. Factors such as individual judges' positions, case overload, and resource shortages emerged as key contributors. We believe that shedding light on these disparities is a crucial step forward to get positive changes within the Spanish justice system.","We utilized a combination of tools and technologies to conduct a comprehensive analysis. R, RStudio, and Excel were employed for data compilation and analysis. Graphics and maps were programmed using D3.js, Canvas, and Javascript. For the creation of straightforward charts, we utilized Datawrapper.","This project shows how the information available in the justice system could be exploited in other countries to check if there are disparities in the decisions of the justice system depending on the court judge on duty. This project shows how detailed data analysis could lead to reveal disparities and inequalities in action of justice, access to health or geographic segregation in every country. 

In Spain, access to highly disaggregated government data is more the exception than the rule. A limitation that often prevents this type of data analysis and the discovery of inequalities in public action. 

The narrative of this investigation can serve as an inspiration for journalists worldwide, encouraging them to proactively request and analyze highly disaggregated data from familiar statistical sources. While regional disparities are acknowledged, this story prompts a critical question: do we truly understand the extent to which inequality multiplies within specific contexts, be it educational centers, healthcare facilities, courts, or administrative centers? This inquiry invites a broader exploration into the nuanced landscape of disparities that may exist beyond regional boundaries, fostering a more comprehensive understanding of societal inequities.",N/A,N/A,Spanish,https://www.eldiario.es/sociedad/espana-protege-forma-desigual-victimas-violencia-genero-juzgados-rechazan-70-ordenes_1_10727651.html,https://www.eldiario.es/sociedad/buscador-ordenes-proteccion-violencia-genero-conceden-juzgado-juzgado_1_10729564.html,,,,,,
HUMANS ARE BIASED. GENERATIVE AI IS EVEN WORSE,Single project,Bloomberg News,Big,"Leonardo Nicoletti, Dina Bass, Chloe Whiteaker, Jillian Ward","Leonardo Nicoletti is a data visualization reporter at Bloomberg News based in New York. He uses data and graphics to tell stories about global issues.

Dina Bass is a technology and artificial intelligence reporter for Bloomberg News based in Seattle, writing about Microsoft and the opportunities and challenges of the rapidly changing AI landscape. 

Chloe Whiteaker is a data visualization editor at Bloomberg News in New York.

Jillian Ward is Bloomberg's managing editor for US technology news on the West Coast, overseeing Big Tech and startup/VC coverage from San Francisco.",United States of America,"Bloomberg conducted a months-long data investigation that found Stable Diffusion—one of the most widely used AI image generators—exacerbates racial and gender stereotypes. We prompted the text-to-image model to create more than 5,000 images of workers with various jobs, as well as people associated with crime, and compared the results with US government data on race and gender. The analysis revealed that not only does the AI model replicate real-world stereotypes—it amplifies them. Through data visualizations and AI-generated photographs, the story explored how these systemic biases in generative AI have the frightening potential to exacerbate real-world inequalities.","This story has been cited and shared extensively – by other media organizations, on social media, and by academic researchers, institutions and governing bodies as important analysis to consider when developing AI safeguards. It spurred serious discussion among AI leaders, catching the attention of Stability AI’s CEO and many other key players in the world of tech, international policy, and AI ethics, like Dr. Buolamwini. 

One startup began developing a generative AI model trained specifically to address racial bias – its founders said our story underscored the need for their approach. Companies like IBM have cited our story in their bias guides. It’s also been cited in reference to AI-policy recommendations for the city of Boston, as well as in official documents about AI risk from the UK government, international bodies such as the United Nations and IMF, and NGOs like Amnesty International. Nicoletti was invited to present findings at the UN HeForShe Summit and the ChangeNOW Summit. The story was cited by several academic research papers and used as a main reference for AI-best practices at dozens of institutions like MIT and University of Toronto. The story was followed by other outlets including the Washington Post and Rest of The World.","Generating thousands of images with a diffusion model can be very expensive if run in the cloud through a third party. As such, we ran Stable Diffusion locally, which is very computationally expensive. To generate our dataset, we ran our script for several weeks.

Another challenge was to make sure that the analysis itself would be as free from bias as possible. Measuring gender and racial bias in images is difficult: before any analysis can be conducted, each image must be labeled with a gender and a skin tone — pinpointing race from an image alone is impossible. With images of real people, this task would be time consuming but straightforward, assuming we could speak to each individual depicted and ask about their identity. The fact that our images were AI-generated added even more complexity – if the subject of an image is not real, how can we accurately label their perceived gender and race? This concern led us to extensive consultations with leading academics, lawyers, and AI ethics experts for a review of the methodology, the use of cutting-edge computer vision techniques, and to the eventual manual review of all 5,100 images by Bloomberg reporters.","Bloomberg’s methods for this experiment were inspired by the groundbreaking work of Joy Buolamwini and Timnit Gebru on the research paper Gender Shades. 

Data collection: Bloomberg used Stable Diffusion v1.5 and Python to generate 5,100 images.

Gender labeling: Given the well-documented inaccuracies of gender classification models, a team of eight reporters reviewed and labeled every image according to the perceived gender of the subject pictured. 

Skin-tone classification: Using Python and the YCbCr algorithm, each image was cropped to the subject’s face and classified which pixels represented skin. Using the average Red Green Blue (RGB) value of skin-pixels, we calculated the Individual Typology Angle (ITA), a statistical dermatology value that relates inversely with the melanin index. ITA was mapped to one of the six Fitzpatrick skin types using an equation developed by Bino and Bernerd. The average skin color was converted from RGB to grayscale to calculate the overall lightness or darkness value on a scale of 0-255.

Average faces: All 300 images generated for a given category were combined using Facer, an open-source Python package.
Graphics: Svelte, D3.js, html <canvas>

A more detailed description of our methods can be found in the methodology section of the story.","Our world is increasingly run by algorithms. Some of these algorithms seem to solve a problem for one group of people, while creating harm for other groups. It’s crucial that algorithms are audited by parties that are working in the best interest of the public. In this context, data-journalists have both the skills and the platforms to audit technology for social good.

This project is an example of the role that data-journalists can play in protecting public interests by (a) conducting algorithmic audits for the purpose of keeping tech companies accountable for their products’ shortcomings, (b) communicating the results of these audits to a non-technical general audience, and (c) influencing real changes across private and public spheres directed at minimizing algorithmic harm.",N/A,N/A,English,https://www.bloomberg.com/graphics/2023-generative-ai-bias/?sref=pmOAaxNp,https://youtu.be/RXp0F9LjAzY?feature=shared&t=2785,https://www.amnestyusa.org/wp-content/uploads/2023/12/Amnesty-VCGenAi-Report-V5.pdf,https://mitsloanedtech.mit.edu/ai/teach/getting-started/,,,,
Inside the Suspicion Machine,Single project,"Lighthouse Reports, WIRED, Vers Beton, OpenRotterdam",Big,"Gabriel Geiger, Eva Constantaras, Justin-Casimir Braun, Htet Aung, Evaline Schot, Saskia Klassen, Romy van Dijk, Dhruv Mehrota, Morgan Meaker, Matthew Burgess, Kyle Thomas, Alyssa Walker, Raagul Nagendran, Hari Moorthy, Ishita Tiwari, Danielle Carrick, Lily Boyce, Andrew Couts, James Temperton, Daniel Howden, Eeva Liukku, David Davidson, Willemijn Sneep, Sascha Meijer,","Lighthouse Reports pioneers collaborative journalism and works with the world’s leading media to deliver deeply reported, public interest investigations. Our core areas of interest are migration, climate, conflict and corruption. We build newsrooms around topics and channel editors, tools and resources to working journalists with the results reaching the public on existing platforms. We have co-published series and investigations that have reached more than 30m people across 100 media partners. We work across media formats from television and documentary to news and radio, podcasts, print and online.",Netherlands,"Suspicion Machines is a four-part series co-published by Lighthouse Reports, WIRED and Rotterdam metro paper Vers Beton. The series unpacks the disturbing scale of automated decision making systems across Europe. The centrepiece of the series, Inside the Suspicion Machine, investigates a machine learning model deployed by Rotterdam to predict which welfare recipients are committing fraud. The story is based on a rigorous statistical analysis we conducted after obtaining unprecedented access to the system’s code and training data. Our investigation revealed that the algorithm is biassed against the young, women, single parents and non-native Dutch speakers, and does not work.","After publication, Rotterdam halted plans to replace the system, which was currently under review, and city council members questioned the mayor. In June 2023, the Dutch Data Protection Authority opened an investigation into the city. 

The investigation was noticed by key figures involved in the European Union’s AI Act, landmark legislation meant to reign in high risk AI technology. Kim van Sparrentak, one of the rapporteurs for the AI sent, sent questions based on our investigation asking how the Act could implement provisions that would prevent Rotterdam-like situations.

The investigation made the case that journalists should redouble efforts to scrutinise black box algorithms deployed by governments. The Markup described the level of access we obtained as a “small miracle.” Algorithm Watch’s Nicholas Keyser Brill described it as“a new milestone in algorithmic accountability reporting” after ProPublica’s landmark Machine Bias investigation that “sets a new professional standard for audits of black-box systems by third-parties.” 

After publication agencies across Europe have decided to share the underlying code of similar systems. We have teamed up with European newsrooms to investigate systems in their own countries, which so far has resulted in major investigations with Le Monde in France and El Confidencial in Spain.","Journalists have struggled to obtain access to the underlying code and data that power predictive machine learning algorithms like the one deployed in Rotterdam. After filing requests in dozens of countries and two years of negotiations, our team was able to use freedom-of-information laws to obtain unprecedented access to the source code, machine learning model and training data of Rotterdam’s algorithm. But this presented an enormous challenge, because journalists had never gained such complete access to be able to audit a public algorithm — we needed to design a methodology from scratch. 

It was important to us that our results were reproducible and our methodology was bulletproof. We convened an eight-person academic panel that reviewed first our experimental design and then our and our results and complete methodology. The panel included experts from the social sciences, computer science, law, and public policy. It also included Jeff Larson, the data journalist behind ProPublica’s landmark Machine Bias investigation. Each expert provided feedback on methods and framing that were incorporated into each step of the experiment.","A unique characteristic of Rotterdam’s machine learning model is that it’s highly non-linear. Variables are not given a single, static ‘weight’ that adds or subtracts from the score. Rather, each variable’s (ie. gender) effect on the scores is dependent on an individual’s unique composition of traits. In other words, variables affect an individual's risk scores in dramatically different ways. 

We wanted to make sure we conducted both controlled and uncontrolled statistical tests. Creating a controlled experiment for a non-linear model proved particularly challenging and there was no journalistic reference point we could look to. We ultimately designed a test that made identical copies of the training data, changed one variable, and then calculated a new risk score for each individual. For gender, for example, we made one copy of the training data with everyone set to female and another with everyone set to male. We then compared scores to determine the average impact of variables. 

Finally, we tested the interaction of multiple variables, allowing us to see how vulnerabilities intersected in ‘profiles’, like single migrant mothers. 

We also analysed the representativeness of the training data and the system’s performance. We produced a fully anonymized synthetic training dataset to ensure reproducibility.","It was a top priority for us that readers know not only what we found, but also understand how we found it. We believe that the public has a vital role to play in algorithmic accountability, but can only fulfil this role if they understand how systems work. The interactive design and use of two archetypes’ journey through the system is designed to educate the public in the choices made when choosing inputs, the vital role of robust training data, the ways complex models obfuscate and legitimise discriminatory technology and the ways that discrimination manifests at scale.

To achieve this, we dedicated a large amount of time and resources to putting together a team that could produce a human-centred narratives, a visual and editorial explainer, and accessible materials to empower other journalists to take on more of this reporting. We needed to build a multidisciplinary team with community trust, which is why we brought together a global media brand, WIRED, with a local newsroom, Vers Beton. 
Reproducibility and transparent methodologies enable accountability. By sharing what we learned, we saw other newsrooms and universities that might have been intimidated in the past decide to take on more of this reporting.",N/A,N/A,"English, Dutch",https://www.wired.com/preview/story/63e565abcde6e0e4f71292cf?status=draft&t=1675978318256,https://www.lighthousereports.com/suspicion-machines-methodology/,https://github.com/Lighthouse-Reports/suspicion_machine,https://www.versbeton.nl/2023/03/computer-zegt-vrouw-hoe-een-rotterdams-algoritme-jonge-alleenstaande-moeders-discrimineerde/?share_code=RfBpxVk5ePHl,,,,
Invisible queues at Social Security: not even a robot can find an appointment,Single project,Civio and El País,Small,"EVA BELMONTE, ADRIÁN MAQUEDA, DAVID CABO, ÁNGELA BERNARDO, CARMEN TORRECILLAS and MARÍA ÁLVAREZ DEL VAYO","Eva Belmonte is co-director and designs, leads, and monitors all of Civio's journalistic investigations. Adrián Maqueda is a data visualization specialist. David Cabo is co-director and and a sofware engineer with an expertise in open data, public data analysis and budget data, among other fields. Ángela Bernardo is deputy director and reporter for health and public policies. Carmen Torrecillas is in charge of frontend development, data visualizations and graphic design for Civio’s projects. María Álvarez del Vayo is a journalist specialized in investigation, data and visualization.",Spain,"This investigation explores the difficulties encountered by individuals in Spain seeking Social Security appointments, specifically for pensions and the Minimum Living Income benefit. Using a robot, we systematically searched for available appointments. Despite consistent efforts over two weeks, our findings revealed that out of 414 offices, 43 lacked appointments for pensions, rising to 63 out of 394 for the Minimum Living Income. Even when appointments were available, half the time they were scheduled at least two weeks later.","Long after covid, some things remain the same in the administration, such as the requirement to make a prior appointment at social security offices. The inability to go to the office for a consultationwithout it was a real problem for people with tight time limits for receiving social benefits. The investigation we presented for this prize showed that in some cities it was almost impossible to get such an prior appointment. However, shortly after publication, the head of Spain's Ministry of Inclusion, Social Security and Migration told, in the Congress of Deputies, that the need for a prior appointment to access social security offices was fake news. We used social media to debunk the claim, showing how the social security website reaffirmed the denial of assistance without an appointment. Two days later, the message on the social security website was removed and the administration was now taking people into the offices without prior appointment. 

The investigation was published simultaneously on the websites of Civio and El País. On the day of publication it had 25,000 readers on each site. It was later republished by Cadena SER, Telemadrid, COPE and in regional media in Asturias, Aragón, Cataluña, Navarra and País Vasco.","Despite the Spanish government's introduction of several social benefits, a considerable number of potential beneficiaries remain underserved. At Civio, we delved into an investigation centering on two critical benefits: the Minimum Living Income, designed to aid low-income families, and pensions. The application process for these benefits is known for its bureaucratic complexity, primarily designed for online submissions with stringent deadlines. Unfortunately, some families lack the means or ability to navigate this process accurately, leading to a failure in accessing their entitled support. In this context, face-to-face administrative appointments can be very useful to fill the gap, so we opted to scrutinize the genuine accessibility of these appointments. Seeking national-level data on appointment availability from public administration would have proven laborious and contingent on the administration possessing such information. So we did our own research: among the 414 offices, 43 lacked available appointments for pension-related matters, and this figure increased to 63 out of 394 for the Minimum Living Income. Additionally, when an appointment was accessible, half the time it was scheduled at least two weeks later.","For this investigation, a Ruby web scraper was used to extract appointment data from the Social Security website across all regions of Spain (17). Data was collected three times a day from 6 March to 19 March, with gaps identified on 13 and 19 March at 17:00. A complete list of social security centres was downloaded to account for missing offices. The resulting data was visualised using D3.js in Observable, showing open offices according to the Social Security list and including five offices with available appointments that were missing from the extraction. Data cleaning in R ensured consistent records for each office. The final visualisation, created using JavaScript and D3.js, depicted delays in appointment availability. The graph ranked offices or provinces from worst to best taking into account delays and lack of availability.","In certain scenarios, journalists face challenges in proving the effectiveness of public administration initiatives. The process of obtaining data directly from these bodies tends to be time-consuming, potentially hindering reporting on the functionality of the policies being implemented. In such cases, web scraping emerges as a viable solution, as illustrated by a recent research project.

At Civio, we pride ourselves on fostering a collaborative environment that seamlessly blends diverse professional profiles. Our team includes individuals with expertise in both journalism and technology, facilitating the intersection of these disciplines. This synergy enables us to efficiently navigate the complexities of information gathering, ensuring that our reporting remains agile and responsive to the dynamic landscape of public policies.",N/A,N/A,Spanish,https://civio.es/2023/03/31/cita-previa/,,,,,,,
Is data neutral ? How an algorithm decides which French households to audit for welfare fraud,Single project,"Lighthouse Reports, Le Monde",Big,"Soizic Penicaud (Lighthouse Reports), Justin-Casimir Braun (Lighthouse Reports), Gabriel Geiger (Lighthouse Reports), Daniel Howden (Lighthouse Reports), Eva Constantaras (Lighthouse Reports), Adrien Sénécat (Le Monde), Manon Romain (Le Monde), Maxime Vaudano (Le Monde), Jonathan Parienté (Le Monde), Elsa Delmas (Le Monde), Léa Girardot (Le Monde), Thomas Steffen (Le Monde)","Adrien Sénécat, Soizic Pénicaud and Gabriel Geiger did research, met with sources and conducted interviews. 
Manon Romain and Justin-Casimir Braun are the data-journalists who took apart the algorithm.
Elsa Delmas, Léa Girardot and Thomas Steffen are part of the design team at Le Monde.
Daniel Howden, Maxime Vaudano, Jonathan Parienté and Eva Constantaras supervised the project.",France,"Le Monde and Lighthouse Reports obtained unprecedented access to the algorithm deployed by the CNAF, the administration responsible for distributing welfare to about half the French population. The ""data-mining"" algorithm is ostensibly used to prioritize fraud investigations, but our work revealed it targets the most vulnerable. We analyzed thousands of lines of code to demonstrate discrimination against parents, low income households, and people with disabilities. On December 4, Le Monde and Lighthouse Reports published an interactive that presented the results of our analysis and the consequences of being flagged for investigation.","Our portrait of a single mom who lost her benefits due to a control triggered by the algorithm was the most visited of our series with more 75,000 unique visitors.

Our publication saw significant secondary coverage, including a column from French economist and author Thomas Piketty that denounced the latent “anti-poor ideology” of the algorithm. Seine-Saint-Denis, the poorest département in mainland France, appealed to the French ombudsman (Défenseure des Droits) to investigate if the algorithm was discriminating against its residents. ATD Fourth World — one of France’s most prominent civil rights organizations — called on the CNAF to halt use of the system. 

In parliament, French MPs called for an investigation and the Minister of the Disabled was forced to answer questions about the CNAF’s use of the system.","Our work has benefited greatly from the efforts of ""La Quadrature du Net"", which has been fighting for years to obtain the code and documents relating to the algorithm. 

The French law very clearly specified that code is a document that can be obtained through FOIA law. But the law is poorly understood and not well enforced. 

Despite the opinion of the French Commission for Access to Administrative Documents (CADA), CNAF still refuses, to this day, to transmit all the variables used by their algorithm, as the CADA's decision is non-binding. We have had to take our case to the administrative court to enforce our right to information, but this will take a year at best.","Having obtained most of the penultimate version of the algorithm, we first had to translate this code from SAS, a language we didn't know, to vanilla Python. Our experience in artificial intelligence enabled us to quickly identify a logistic regression, a very standard algorithm. Each variable could then be taken separately, and its impact analyzed for each of the possible values. We transferred all these coefficients to a Google spreadsheet, so that each member of the project could play around with the model.","Analyzing algorithms can be intimidating. The administrations that use them often exaggerate their technicality and hide the actual code at all costs. Authorities are often reluctant to accept that their algorithms, although mathematically correct, are not neutral. They often refuse to conduct bias studies on the pretext that they would be useless, and yet deploy these solutions in every aspect of a citizen's life.",N/A,N/A,"French,English",https://www.lemonde.fr/en/les-decodeurs/visuel/2023/12/05/how-an-algorithm-decides-which-french-households-to-audit-for-benefit-fraud_6313254_8.html,https://www.lighthousereports.com/methodology/how-we-investigated-frances-mass-profiling-machine/,https://www.lemonde.fr/les-decodeurs/article/2023/12/04/dans-la-vie-de-juliette-mere-isolee-precaire-et-cible-de-l-algorithme-des-caf_6203803_4355770.html,,,,,
Know their names: Palestinians killed in Israeli attacks on Gaza,Single project,Al Jazeera Digital,Big,"Mohammed Haddad, Mohammed Hussein and Konstantinos Antonopoulos",.,Qatar,"“We are not numbers” - those are the words repeatedly uttered by Palestinians in Gaza who have been living under constant Israeli bombardment for over 100 days.

As data journalists at AJLabs, our mission has always been to tell the human stories behind the data. As of writing this, more than 24,000 people have been killed in Israeli attacks on Gaza. These are people with stories, dreams and hopes for the future. 

Know their names is a data/visual story, with a searchable database, that documents the thousands of people that have been killed during Israel’s most recent military campaign.","This story, which was published in English, Arabic and Bosnian has been one of the most widely circulated stories on the continuing Israel-Gaza war.

Shared online, on screen, on posters and in protests around the world, the list of names of those killed, more than 70 percent of them being women and children, continues to provide the human stories of those living in Gaza.",Producing a database of names and ages is not what made this a compelling data-driven story. Rather it’s through the nuanced presentation of the human stories weaved within the visuals to help people understand the scale of the attacks.,"Given this was a breaking news story, speed played a huge part in publishing this story. It took a team of three people three days to storyboard, translate, analyze, produce and edit the story.

Some of the main tools included:
ChatGPT, Google Translate and a home grown AI-assistive tool to run the first pass of translations. These were then manually verified to ensure that all the names were accurate.
With our cleaned spreadsheet, we then used the Adobe Suite of tools including Illustrator, Premiere and After Effects to produce all the visual assets.
Finally, we used DataWrapper to create the searchable database.",Always strive to tell the human stories behind the data.,N/A,N/A,"English, Arabic, Bosnian",https://www.aljazeera.com/news/longform/2023/11/1/know-their-names-palestinians-killed-in-israeli-attacks-on-gaza,https://www.aljazeera.net/knowledgegate/newscoverage/longform/2023/11/8/%D8%A8%D8%A3%D8%B3%D9%85%D8%A7%D8%A6%D9%87%D9%85-%D9%88%D8%A3%D8%B9%D9%85%D8%A7%D8%B1%D9%87%D9%85-%D8%B7%D9%84%D9%8A%D8%B9%D8%A9-%D9%82%D8%A7%D9%81%D9%84%D8%A9-%D8%B4%D9%87%D8%AF%D8%A7%D8%A1,https://www.aljazeera.com/news/longform/2023/12/12/know-their-names-palestinians-killed-by-israel-in-the-occupied-west-bank-2,https://balkans.aljazeera.net/interactives/longform/2023/11/2/pamtite-njihova-imena,,,,
L.A. Homelessness: How LA’s Scoring System for Subsidized Housing Gives Black and Latino People Experiencing Homelessness Lower Priority Scores,Single project,The Markup,Small,"Colin Lecher
Maddy Varner","Colin Lecher is a reporter based in New York. His work on technology and society has received honors from the Association of Health Care Journalists, the Deadline Club, the Radio Television Digital News Association, and more. Before joining The Markup, he was a senior reporter with The Verge.
Maddy Varner extracts datasets from public documents and databases to help build a more concrete understanding of how organizations use technology to predict and affect behavior—and the problems in those assumptions. Before The Markup, she was a researcher at ProPublica, where her team won a Loeb Award for Beat Reporting in 2017.",United States of America,"A Markup investigation, also published by the Los Angeles Times, confirmed what advocates for the unhoused had long suspected: For years, the scoring system for allocating housing on the basis of vulnerability rated unhoused Black people as less vulnerable than White people and, as a result, deprioritized their candidacy for permanent housing. 

The Markup was the first organization to obtain breakdowns of more than 130,000 “vulnerability” scores assigned to unhoused people in LA, going back to 2016. Our data analysis found a persistent discrepancy in scores between Black and White people experiencing homelessness.","Our article helped lift the deliberate veil of secrecy around this scoring tool, making public the questions asked and the scoring system itself—previously, case workers were specifically instructed not to tell unhoused people they were being scored.

Our published stories included tips for how unhoused people could get re-scored and what criteria case managers consider when deciding whether to re-score someone. We also partnered with the Los Angeles Times to help ensure wide distribution in the communities that care about and are impacted by the scoring system.

After publication, sources told us that specifically because of our reporting, Los Angeles city regulators are making progress on a way to address the problem.","For years, people have known that the scoring system used in Los Angeles to decide who receives subsidized housing likely had troubling racial disparities. But the evidence remained largely anecdotal. 

In August 2021, a Markup reporter in New York, Colin Lecher set out to change that and sent his first public records request to the Los Angeles Homeless Services Authority (LAHSA), asking what scores people received, what help they ultimately received (if any), and whether race played a factor in how people were scored and got help. 

Over the next 18 months, Colin teamed up with Markup data reporter Maddy Varner, and after more than a year of back and forth with the authority, became the first to obtain 130,000 scoring surveys going back to 2016. 

The resulting investigation confirmed that the system had for years rated unhoused Black people as less vulnerable than White people, giving them lower priority for permanent housing. 

In discussions with the Los Angeles Times, their lead homelessness reporter said that this was a story that they had been wanting to do for years, and the Times ran Colin and Maddy’s story on the front page of the paper.","A substantial portion of the data work on this project involved persistently seeking information from LAHSA (Los Angeles Homeless Services Authority). It took several months of back-and-forth with the agency to identify which of LAHSA’s databases held the relevant data, and several more months to figure out the actual definition of the data’s labels and categories.
 
The data we used came from vulnerability assessments of people experiencing homelessness. We sorted individuals into one of three groups based on their score (also categorized as “low,” “mid,” “high”) as the agency does. We then analyzed what percentage of White, Black, and Latino adults were categorized in the high vulnerability group.

To determine if the differences in those percentages were statistically significant, we ran a chi-square analysis of vulnerability groups by race. To check whether these disparities might be explained by other variables in the dataset, we performed a series of logistic and linear regressions. We also conducted a linear regression with the total vulnerability score as the dependent variable as well as a logistic regression in which the dependent variable was whether someone’s score put them in the high acuity group. For full details, see “How We Investigated L.A.’s Homelessness Scoring System.”","1. Find as much documentation as possible

Although the data we were looking for wasn’t publicly available, we found blank surveys on the website of the relevant government agency, which gave us details to ask for when we crafted our public records request.

2. After putting in a request, don’t fear the follow‑up

When we first started reporting, we filed a simple data request that asked for a fraction of the data that we ultimately ended up using in our analysis. The more we reported, the more detailed our questions—and requests—became. By the end of our investigation, we followed up to ask for more records at least a half-dozen times, either formally or informally

3. When in doubt, ask the people who use the data every day

We spoke with people who interacted with the scoring system at the center of our story: case managers who administered the surveys, matchers who used the results to determine housing placements, and the people who were surveyed themselves. They all helped answer our questions about the data, since they were experts in its nuances and shortcomings. 

See the how-to article for journalists doing similar projects: “[Journalists: Investigate Homeless Vulnerability Scoring in Your City](https://themarkup.org/story-recipes/2023/02/28/journalists-investigate-homeless-vulnerability-scoring-in-your-city).”",N/A,N/A,English,https://themarkup.org/investigation/2023/02/28/l-a-s-scoring-system-for-subsidized-housing-gives-black-and-latino-people-experiencing-homelessness-lower-priority-scores,https://themarkup.org/show-your-work/2023/02/28/how-we-investigated-l-a-s-homelessness-scoring-system,https://themarkup.org/story-recipes/2023/02/28/journalists-investigate-homeless-vulnerability-scoring-in-your-city,,,,,
Looted Forests,Single project,"Le Monde,InfoCongo",Big,"Authors: Josiane Kouagheu and Madeleine Ngeunga
InfoCongo team: David Akana, Kevin Nfor, Jeannot Ema’a, Fabrice Wekak, Akira Junior and Ella Iradukunda
Le Monde team: Charlotte Bozonnet, Marie De Vergès, Elise Barthet, Anna Sylvestre Treiner, Riccardo Pravettoni and the graphic design team
And the data architect of Rainforest Investigations Network: Kuang Keng Kuek Ser","This project was built on the collaboration between a local media outlet, InfoCongo, and an international one, Le Monde. It was a true team effort that involved at least 15 people. Madeleine Ngeunga and Josiane Kouagheu, led the investigation and received guidance from InfoCongo and Le Monde editors as well as from the Rainforest Investigations Network/Pulitzer Center team. The investigation involved data analysts, graphic designers, and illustrators who helped analyze and visualize the data. The team also worked closely with a local reporter who had knowledge and access to the communities.",Cameroon,"This project is a collaboration between a local media outlet, InfoCongo, and an international one, Le Monde, revealing an unprecedented overview of the scale and modus operandi of the illegal timber trade in Cameroon. During 12 months, the two newsrooms analyzed six years' reports on illegal logging and met dozens of timber traffickers, local residents and truck drivers in Cameroon. The four part series showed how powerful individuals use their network to cut, transport and launder timber, from the forest areas to the port and the international market (European Union, China and Vietnam).","This project is a collaboration between a local media outlet, InfoCongo, and an international one, Le Monde, revealing an unprecedented overview of the scale and modus operandi of the illegal timber trade in Cameroon. During 12 months, the two newsrooms analyzed six years' reports on illegal logging and met dozens of timber traffickers, local residents and truck drivers in Cameroon. The four part series showed how powerful individuals use their network to cut, transport and launder timber, from the forest areas to the port and the international market (European Union, China and Vietnam).","The two female journalists who led the project live and work in Cameroon, the second-largest forest country of the Congo Basin, where journalists are threatened, imprisoned, tortured/killed for their work. 
While we were investigating, the dead and tortured body of our colleague Martinez Zogo was found in a township in Yaoundé. He was investigating corruption involving businessmen and public officials. 
Alongside this context, Cameroon, governed by a 90-year-old president who has been in power for over 40 years, is struggling with its forest governance. 
Quarterly, the Ministry of Forests publishes the ""sommier des infractions"", a detailed report on environmental crimes, listing individuals or companies sanctioned for forest law violations. But these reports are not available online, and the Ministry almost never responds to requests for access to the data. From various sources and drawn on our own databases, built up over the years, we compiled six editions of the ""sommier des infractions"", published between 2015 and 2021. 
We also built trust with on the ground sources, traffickers, and truck drivers to investigate how the illegal timber is negotiated with local communities, then transported from the forest to the international market.","The series goes on to contextualize the scale of this illegal trade and connects us with data. The graphics serve as elegant and informative support by summarizing the amount of data processed, revealing which companies are responsible for the concessions and the deforestation. 

As mentioned above, this series analyzed six editions of the ""sommier des infractions"" (quarterly documents listing companies and individuals sanctioned for forest law violations), published between 2015 and 2021, using google sheet. 

We also drew on open data and used tools such as QGIS, to analyze the Cameroon Forest Atlas data, cross-reference with data from the sommier des infractions, to identify the role of legal companies in illegal logging. QGIS also helped us to map the logging permits granted by the Ministry of Forests from 2015 to 2021. Our analysis revealed that companies sanctioned for illegal logging, according to the ""sommier des infractions"" have obtained the largest areas of logging permits in Cameroon over the last six years.
We also used Planet imagery to show forest change over years in areas where timber was illegally harvested. Among other tools, we used datawrapper, flourish to visualize our data.","Trust is our keyword. 

The trust collaboration, newsroom included, is the key. To produce a collaborative work like ours, journalists need to be willing to work together and share everything. 
For an investigation on illegal logging, going on the field is important. But to do it, it is very important to prepare all the aspects, from people to meet to the security of the team. That last point is essential. To do it, having a trusted local journalist/source can help. 

After the testimonies of the populations, sources on the field, it is important to cross-reference with accurate documents/data/sources that can prove the denunciations, accusations or information. 

Then, as journalists, we need to read a lot of reports from the government, NGO or other organizations. Journalists also need to have a strong network of trusted sources and experts that could share information/data with them in a secure way. 

It is also essential to build a trusted relationship with sources involved in illegal practices and carefully listen to them. This helped us to get enough information on the global system of illegal timber trafficking and humanize the sources rather than demonize them.",N/A,N/A,"English,French",https://infocongo.org/en/cameroons-undeterred-illegal-loggers/,https://infocongo.org/en/how-illegal-wood-escapes-control-circuits-in-cameroon/,https://infocongo.org/en/forest-communities-complicity-with-illegal-loggers/,https://infocongo.org/en/caught-between-conflicting-interests-of-govt-and-business-forest-communities-in-cameroon-squeezed-out/,,,,
Missing in Chicago,Single project,"Invisible Institute, City Bureau",Small,"Trina Reynolds-Tyler, Sarah Conway","Trina Reynolds-Tyler, Director of Beneath the Surface at the Invisible Institute, and Sarah Conway, Senior Reporter at City Bureau, partnered to investigate patterns of how police respond to missing Black women and girls in Chicago. The Chicago Missing Persons project is the latest in a long history of collaboration between City Bureau and the Invisible Institute, who both share a foundational ethos of cross-disciplinary, collaborative reporting that reflects community priorities.",United States of America,"The search for missing persons is one a core function of modern police departments. Some cases, like the recent disappearance of Gabby Petito, capture overwhelming media attention, while the hunt for missing persons is a staple of TV shows. While these shows often emphasize stories of police solving cold cases, the reality is often starkly different in Chicago, the third-largest city in the U.S., as our two-year data-based investigation shows based on groundbreaking machine learning and community engagement techniques. Journalists found deep racial disparities and heartbreaking individual stories hidden within the data.","In the two months since the investigation was published and website launched, reporters Trina Reynolds-Tyler and Sarah Conway have testified in front of a task force created by the state legislature to study the issue of Missing and Murdered Women in Chicago, presenting on their findings for over and hour and then taking questions from state representatives. Chicago Mayor Brandon Johnson’s staff has also reached out about inviting both reporters to testify in front of Chicago City Council in the coming weeks. Additionally, the investigation reached a broad audience of Chicagoans through the Chicago Reader’s 60,000-copy print run and through a dedicated mini-site, which garnered over 4,300 unique visitors. Finally, the reporting made a direct impact on policymakers who bear responsibility for Chicago’s troubled handling of missing persons cases. Former CPD superintendent Eddie Johnson apologized for his handling of a prominent missing persons case where he linked the disappearance of two young Black women to their alleged involvement in narcotics. Both were eventually found murdered. “Just [because of] who they are or where they come from or their lifestyle, that doesn’t mean the police shouldn’t take it as serious, because we should,” he told our team. “Somebody loved that person.”","Of particular relevance to the Sigma Awards, this investigation grew out of Beneath The Surface (BTS), a groundbreaking data journalism project of the Invisible Institute, headed by director of data Trina Reynolds-Tyler. To develop BTS, Trina partnered with the Human Rights Data Analysis Group to build a machine learning model that analyzed 30,000 Chicago police complaint files and identified buried patterns of misconduct, including a hidden cluster of cases centering around the mishandling of missing persons cases. Our team has years of experience working with CPD data, but the BTS model’s approach unearthed sets of cases we had never seen before.

Our team also built consistent community engagement into our data process and our reporting, holding five major community events and engaging 200 volunteers to review and classify complaints in order to train the machine learning model. This innovative work directly informed many findings in the series about the poor quality of CPD’s data, and led reporters to multiple anecdotal cases explored in the series. Further information can be found at https://www.btsurface.com/.","To access the police misconduct records, our documents pipeline team used OCR software to scrape the narrative text from specific report types. After gaining access to the documents we utilized R and python is cleaning the data and building the classifier. When training volunteers to label the training data for our classifier, we utilized google sheets. 

Our technique for training volunteers leaned heavily on restorative justice practitioners toolkits. We established community agreements and cultivated a space to discuss heavy topics. We followed a strict agenda with space to check in at the beginning of each session, time apart to label data, and then allocated approximately 30 mins to share out. Our hands-on approach with volunteers also meant that their feedback helped inform the data reporting, helping the team refine tags and identify new topics.","The work of Beneath the Surface to unearth stories hidden within administrative data, obscured by difficult-to-pinpoint decisions that serve the furtherance of police impunity, can directly inform the work of other journalists using public data. In particular, the team demonstrated a clear use of machine learning for sourcing stories from complicated and text-heavy datasets where relying on official categories obscures important trends. Second, the team’s insistence on building community input into the data process is a model for journalists using ML or data reporting. That input strengthened the data journalism, especially by providing a sounding board of community members who had lived experience.

The role of FOIA in that data validation can also inform others. Rather than end at one request that produces a dataset, journalists should look for ways to validate their findings by obtaining a mass of individual case records and interrogating how the “truth” contained in the narratives and other documents conflicts with the “truth” in the raw data. That process is how we were able to make the finding that official data showing that 99.8% of missing persons cases between 2000-2021 were closed as “non-criminal” obscures hundreds of missing persons cases that were improperly classified.",N/A,N/A,English,https://chicagomissingpersons.com/,https://chicagoreader.com/news-politics/chicago-police-missing-persons/,https://thetriibe.com/2023/11/families-of-the-missing-speak-out-about-stigma-police-and-closure/,https://southsideweekly.com/who-goes-missing/,,,,
NarcoFiles: The New Criminal Order,Single project,"Media Partners: Agencia Ocote (Guatemala), Aristegui Noticias (Mexico), Armando.Info (Venezuela), Berlingske (Denmark), BIRD (Bulgaria), Centro Latinoamericano de Investigación Periodística (CLIP), Cerosetenta / 070 (Colombia), CNN en Español (United States), Con Criterio (Guatemala), ContracorrienteHN (Honduras), Cuestión Pública (Colombia), De Tijd (Bélgica), Der Standard (Austria), Die Dunkelkammer (Austria), El Universal (México), Expresso (Portugal), Frontstory.pl (Poland), Het Parool (Nederlands), Infolibre (España), InSight Crime (Latin America), Investigace.cz (Czech Republic), IrpiMedia (Italy), Knack (Belgium), La Prensa (Panamá), Mexicanos contra la Corrupción (México), Miami Herald (US), Mongabay Latam (Latin America), Narcodiario (España), No Ficción (Guatemala), Ojoconmipisto (Guatemala), OjoPúblico (Perú), Paper Trail Media (Germany), piauí (Brazil), Plan V (Ecuador), Plaza Pública (Guatemala), Profil (Austria), Quinto Elemento Lab (Mexico), Siena (Lithuania), SVT (Sweden), Univision (United States), UOL (Brazil), Verdad Abierta (Colombia), Vorágine (Colombia) and ZDF (Germany).",Big,"Journalists: Allan de Abreu (piauí), Samuel Aguilar (El Universal), Iván Alamillo (Mexicanos Contra la Corrupción y la Impunidad), Johanna Álvarez (Unidad Investigativa de Noticias Caracol), Tatiana Antolínez (Cuestión Pública), Carmen Aristegui (Aristegui Noticias), Ernesto Aroche (El Universal), Anastasia Austin (Insight Crime), Claudia Báez (Cuestión Pública), Mayra Báez (CLIP), Valeria Báez (Cuestión Pública), Antonio Baquero (OCCRP), Aldo Benitez (La Volanta), Lars Bové (De Tijd), Scott Bronstein (La Prensa), Edier Buitrago (Cuestión Pública), Àngela Cantador (CLIP), Abel Cárdenas (Ojo Público), Rigoberto Carvajal (CLIP), Brecht Castel (Knack), Aramis Castro (OjoPúblico), Daniela Castro (OCCRP), Jacqueline Charles (Miami Herald), Rosa Chávez Yacila (OjoPúblico), Alejandra Crail (El Universal), Juliana Dal Piva (CLIP), Antonio Delgado (Miami Herald/El Nuevo Herald), Douwe den Held (Insight Crime), Claudia Julieta Duque (Cuestión Pública), David Espino (El Universal), Juan Omar Fierro (Aristegui Noticias), Brian Fitzpatrick (OCCRP), Casey Frank (Miami Herald), Marina Gama Cubas (CLIP), Enrique García (Ojoconmipisto), Jody García (Plaza Pública), David González (OCCRP), Eduardo Goulart (OCCRP), José Guarnizo (Vorágine), Daniela Guazo (El Universal), Kevin G. Hall (OCCRP), Pavla Holcova (investigace.cz/OCCRP), Gianfranco Huamán (OjoPúblico), Nathan Jaccard (OCCRP), Mike LaSusa (InSight Crime), Sol Lauría (La Prensa), Luisa Fernanda López (CLIP), Nelly Luna Amancio (OjoPúblico), Michael Lund (Berlingske), Vinicius Madureira (OCCRP), José Miguel Marulanda (Cuestión Pública), Paul May (Investigace.cz), Angélica Medinilla (Agencia Ocote), Stefan Melichar (Profil), Víctor Méndez (Narcodiario), Silber Meza (El Universal), Anastasiia Morozova (Frontstory.pl), Hannes Münzinger (Paper Trail), Belisa Murillo, Michael Nikbakhsh (Die Dunkelkammer), Raúl Olmos (Mexicanos Contra la Corrupción y la Impunidad), Bianca Padró Ocasio (OjoPúblico), Angus Peacock (OCCRP), José Luis Peñarredonda (CLIP), Micael Pereira (Expresso), Paul Radu (OCCRP), Begoña P. Ramírez (InfoLibre), Miriam Ramírez (El Universal), Gerardo Reyes (Univisión), Andrea Rincón (Cuestión Pública), María Teresa Ronderos (CLIP), Ketty Roqueme (Cuestión Pública), Iván Ruiz (CLIP), Nicolás Sánchez (Vorágine), Violeta Santiago (Quinto Elemento Lab), Lilia Saúl (OCCRP), Mariusz Sepioło (Frontstory.pl), Khadija Sharife (OCCRP), Dimitar Stoyanov (BIRD), Konrad Szczygieł (Frontstory.pl), Armando Talamantes (Quinto Elemento Lab), Atanas Tchobanov (BIRD), Juanita Vélez (CLIP-OCCRP), Laura Weffer (OCCRP), Jonny Wrate (OCCRP)

Editing: Sally Mairs (OCCRP), Alexander Dziadosz (OCCRP), Jared Ferrie (OCCRP), Miranda Patrucic (OCCRP)

Research, Data and Tech: Ignacia Velasco (OCCRP), Jan Strozyk (OCCRP), Eric Barrett (OCCRP), Angus Peacock (OCCRP), Sharad Vyas (OCCRP), Benjamin Spahovic (OCCRP), Nedim Cengic (OCCRP)

Fact-Checking: Birgit Brauer (OCCRP), Ivana Jeremić (OCCRP), Olena LaFoy (OCCRP), Bojana Pavlović (OCCRP), Maura Quatorze (OCCRP)

Promotion: Sergiu Brega (OCCRP), Kathlyn Clore (OCCRP), Lauren Jackman (OCCRP)

Design and Graphics: James O'Brien (OCCRP), Edin Pašović (OCCRP)

Web and Interactive: Alex de Santa (OCCRP)

Project Coordination: OCCRP

Main partner: CLIP","OCCRP’s Latin American team lead NarcoFiles. It includes: Nathan Jaccard (global coordination - edition), Daniela Castro (coordination - edition), Antonio Baquero (coordination), Ignacia Velasco (data - research), Lilia Rodríguez (coordination Mexico), David Gonzalez (security - edition), Kevin G. Hall (coordination - edition), Angus Peacock (research) 
…..
Nathan Jaccard is the Latin America editor at the Organized Crime and Corruption Reporting Project since 2018. He coordinates cross-border investigations in the region and brings local partners into global projects.",Netherlands,"NarcoFiles is the largest investigation on organized crime to originate in Latin America. The project, sparked by a leak of emails from the Colombian prosecutor’s office, allowed OCCRP and 40+ media outlets to look into how criminal gangs are innovating and evolving their business models in the face of new opportunities to feed one of the world’s busiest drug highways: Latin America to Europe. 

This collaboration posed unprecedented data challenges that had to be solved in order to ensure the security of the transnational team and the integrity of the data.","80+ media outlets from around the world republished the stories from the project in different languages, giving the stories widespread visibility. The NarcoFiles project was one of the most visited in the history of the OCCRP website. The project team will continue to investigate the findings in the leak and more stories will be published throughout the year. 

NarcoFiles represented a significant milestone in investigative journalism: it is the first global cross border collaboration originating from Latin America focused on organized crime and a keen example of what can be done in the region. The network created by NarcoFiles is a seed for further Latin America-focused collaborative investigations.

Colombia's President, Gustavo Petro, recognized the project during the ceremony for the 132nd anniversary of the National Police, explicitly mentioning the work carried out by Colombian outlets such as Vorágine and Cuestión Pública. 

Journalists who were involved in NarcoFiles conducted a workshop at the Latin American Conference of Investigative Journalism (COLPIN) in Mexico in December 2023. This opportunity to share knowledge & experiences with other media professionals showcased the regional influence and the international implications. NarcoFiles was covered by a range of media outlets, including a VICE interview and the Latam Journalism Review.","The sensitive nature of the data made the project one of the most complex in OCCRP's history. The size of the database - almost double that of the Panama Papers - required meticulous coordination between different teams at OCCRP and our partners. In addition, our team also tailored meetings and responses for the most complex stories / regions, customized digital and physical security protocols for the most sensitive stories, centralized and coordinated more than 40 FOIA requests and rights of reply (more than one dozen) so as not to compromise local journalists.

The political context was also challenging. The current Attorney General's Office is not transparent and did not address the findings of the NarcoFiles investigation. The NarcoFiles exposed the seriousness of the attack and the sensitivity of the data that was compromised. In addition, the Attorney General’s Office attacked some journalists for disclosing information of public interest.","OCCRP, the Centro Latinoamericano de Investigación Periodística (CLIP), Vorágine, and Cerosetenta / 070 gained early access to 5tb of data from two organizations, Distributed Denial of Secrets and Enlace Hacktivista. The dataset, with more than seven million emails from the Colombian prosecutor’s office, included audio clips, PDFs, spreadsheets, and calendars, dated to 2001. OCCRP collated and cleaned the data, getting rid of problematic content, which was tedious process. We identified patterns and did a lot of number crunching on the backend across the dataset to give the investigative team the necessary leads to follow. OCCRP brought in 40+ media organizations from 23 countries across the Americas and Europe to analyze the contents. The team also filed more 40 FOI requests in 21 countries and conducted data analysis on the responses, which in combination with interviews with sources, and hundreds of public and private documents, allowed OCCRP and partners to build compelling interactives to show how organized crime has become more globalized than ever. For the interactive, our team used 3D (Blender), Gridsome (web frontend framework base on Vue.js) and GSAP for animations.","The lessons from NarcoFiles offer valuable insights for other journalists. Managing a massive leak from Colombia's Public Prosecutor's Office, with 5 terabytes of data, required technological infrastructure for secure data storage and efficient navigation to sift through this sea of information to find what truly mattered to the public. Effective use of OCCRP's data platform, Aleph, proved vital in this challenging process.

It was also a challenge to confirm the authenticity of the information in the leak. Reporters had to cross-reference court data, public records, and other databases. They also submitted public information requests, reviewed documents, and interviewed experts.

The sensitive nature of NarcoFiles made necessary security strategies, both digital and physical, to mitigate risks of violence against involved journalists. Developing protocols was also critical for protection of undercover agents and informants within the leaks.

Coordinating over 40 media outlets worldwide posed a monumental challenge, including overcoming cultural differences and language barriers, especially with non-Spanish speaking countries, underscoring the need for effective management strategies and seamless communication to ensure collaboration in such complex transnational journalistic projects. These learnings emphasize the importance of security, ethics, strategic collaboration, and efficient technological use in conducting large-scale journalistic investigations.",N/A,N/A,"English, Spanish, Danish, Bulgarian, German, Dutch, Czech, Polish, Swedish, Lithuanian, French, Portuguese",https://www.occrp.org/en/narcofiles-the-new-criminal-order/,https://latamjournalismreview.org/articles/starting-in-latin-america-narcofiles-brings-together-journalists-from-23-countries-to-reveal-new-global-organized-crime-networks/,https://www.vice.com/en/article/pkax7n/narcofiles-papers,https://www.youtube.com/watch?v=HrBe46R-r1s,,,,
Predators in The Classrooms,Single project,"El Universal, Connectas.",Big,"Daniela Guazo, Alejandra Crail, Daniel Gómez, Ani Cortés, Liliana Pedraza, Paola Reyes, Atenea Campuzano","El Universal is the only newspaper in Mexico that has an Investigative Journalism and Data Unit made up of investigative journalists, data journalists and programmers worked together on this kind of projects. The team for this:
**Daniela Guazo Manzo**. Coordinator of the Data Unit.For the past six years she has been teaching workshops in different parts of Mexico on data analysis.
**Alejandra Crail** is an investigative journalist specializing in corruption, human rights, children and gender. 
**Daniel Gómez Hernández**, physicist by the UNAM. He oriented his studies towards programming, data analysis and complex systems to understand the situation of his environment.",Mexico,"Predators in The Classrooms' is the first project in Mexico that has a database with more than three thousand sexual assaults against students of pre-school, elementary school and middle school throughout Mexico in the last decade by school workers.

Through more than 500 requests for transparency addressed to the Ministry of Education, the National Human Rights Commission and the prosecutors' offices in the 32 states, we were able to identify the profile of almost 100 aggressors. 

We built a tool that allow to review the protocols published by the 32 states of Mexico to combat sexual violence at the schools.","The database was the most current information on child abuse perpetrated by school workers in Mexico. It is the first time that journalists offer a complete landscape of the problem of elementary school and middle school to the audience. 

After the publication, Congress approved a ruling to modify the General Law on Women's Access to a Life Free of Violence in order to grant the Ministry of Public Education new faculties to develop and apply protocols for the prevention, attention, punishment and eradication of gender violence at all educational levels. In Mexico City, Congressman Carlos Joaquin Fernandez met with Daniela Guazo and Alejandra Crail in order to learn about the investigation in detail and to promote initiatives in favor of the victims of sexual abuse in schools. Civil society organizations have taken the work as a basis for academic research on this issue. Likewise, the teaching community has held virtual dialogue tables to address this hidden reality in the country's schools.","The number of cases of abuses perpetrated by school workers reported by the press has increased in recent years. These public alerts have shown the inefficiency of the school authorities, who had been protecting the abuses deliberately or as a way of protecting themselves. Despite of this, authorities deny constantly that this is the reality in the schools of Mexico.

One of the biggest challenges of this project was to obtain the data. In Mexico 32 different institutes process the information in their way. The result was that most of the scholar authorities tried to block access to the complaints against school workers. Some institutes didn't give the data, others gave information in images and we had to read several complaints written by students, their parents or other school workers to obtain more data.

We had to made a strategy to avoid the blocking of the information and at least have all the statistics about this felony. 

We also had a big challenge in terms of testimonies. All the families were afraid for their childrens, so we teamed up with a group of women lawyers who are in charge of child protection.","The Nuxt.js javascript framework was used to develop the microsite. The home section displaying the testimonial documents in 3D was achieved with the Three.js library, while the data visualizations were developed with the D3.js library. The animations and transitions binding to scrolling use the gsap.js library. All the libraries in our project are open source software. You can check our code here: https://github.com/jdanielgoh/abuso-escuelas/","Create a strategy to avoid information blocking, request the same information, but in different ways. This will allow you to create your own database, but with official statistics. 

Have a multidisciplinary team that focuses on innovative formats to tell stories that get lost in the daily information and think in all formats for their investigations. 

A relevant point for us was to make an alliance with civil society specialists in the subject, they helped us with testimonies, to share the project with other groups and we even shared information.",N/A,N/A,Spanish,https://interactivos.eluniversal.com.mx/2023/abuso-sexual-escuelas/,https://interactivos.eluniversal.com.mx/2023/abuso-sexual-escuelas/secreto-escolar,https://interactivos.eluniversal.com.mx/2023/abuso-sexual-escuelas/nuestros-protocolos-no-existen,https://www.youtube.com/watch?v=3B61wcXMq9Y&t=306s,,,,
Roots of an Outbreak,Single project,ProPublica,Big,"Caroline Chen, Irena Hwang, Al Shaw","Caroline Chen covers health care for ProPublica. She has written about public health, hospitals, drugmakers and clinical trials, highlighting disparities in patient access, broken funding models and abuses of power.
Irena Hwang is a data reporter at ProPublica. She has a master’s degree in journalism and a doctorate in electrical engineering from Stanford University, and studied electrical engineering at the Massachusetts Institute of Technology.

Al Shaw is a deputy editor on the news apps team at ProPublica. He uses data and interactive graphics to cover environmental issues, natural disasters and politics.",United States of America,"ProPublica wondered what governments and international health agencies were doing to prevent the next pandemic. In this series we exposed an overlooked cause of outbreaks: deforestation. We applied real world data to a theoretical model, examining sites of previous Ebola outbreaks and showing how they are at higher risk than they were when their outbreaks occurred.

We trained a machine-learning model to determine if places where Ebola had never emerged had similar characteristics to outbreak locations. The model considered millions of locations in 27 countries, and identified 51 locations with patterns of tree loss similar to the previous Ebola outbreaks.","We blended on-the-ground reporting and photography in West Africa with statistical modeling and an original analysis of satellite imagery to examine risky patterns of forest loss. 

It’s easy for Western readers to blame local residents — to blithely say, just stop cutting down the forest. That’s why it was important to let readers hear from residents of Meliandou, the village where the world’s worst Ebola outbreak began. They told us that they have no choice but to keep chipping away at the forest in order to grow food or they’d starve. We found farming experts with ideas that could help villagers increase their yields and break the pattern of constant deforestation. The world has produced more than 40 reports on what went wrong during that epidemic, yet a top Guinean rice expert we interviewed said none of the authors of those reports ever asked him or his colleagues for advice. A story we reported from Madagascar detailed more solutions that few global-health authorities had considered.

Our machine-learning model identified 27 locations concentrated in Nigeria, a country that has never seeded an Ebola outbreak. Reporters found a lack of outbreak preparedness there – spotlighting areas that could benefit from funding and attention.","We traveled to Meliandou, Guinea, to show how this tiny village surrounded by forest is emblematic of the short-sighted global public health system. Meliandou is where a virus that once lived inside a bat found its way into the cells of a toddler, setting off the world’s worst Ebola outbreak. Health care workers clad head to toe in protective gear rushed to West Africa to treat the sick and extinguish the epidemic, an effort that took more than two years and cost at least $3.6 billion. Then, the foreign doctors packed up and the medical tents came down.

We wondered what the world had done to keep disaster from striking again. Reporting in Meliandou, we found that if Ebola were to emerge from the forest again, there’s little to stop the virus from ravaging this village and spreading beyond its borders once more. Residents there still rely on a clinic with no electricity or running water. Patients wash their hands in the same bucket. It’s staffed by a single midwife who has none of the head-to-toe PPE she’d need. “We are suffering,” village chief Jiba Masandouno told us. “The government has forgotten us. The international community has forgotten us.”","We brought the science to life by taking a theoretical mathematical model and applying real world data to it, examining sites of previous Ebola outbreaks and showing how they had changed. To get a sense of the current risk of spillover from deforestation in Meliandou and other sites of Ebola outbreaks, ProPublica consulted with a dozen researchers for its own analysis. We applied data on tree loss from historical satellite images taken between 2000 and 2021 — the most recent year that was available at that time — and tested tens of thousands of infection scenarios. That’s how we determined that sites of prior outbreaks are at higher risk than before.

As a companion piece, we created an interactive page that guided readers through the science, allowing them to cut down an imaginary forest and see how creating more and more patches increases the opportunities for humans and wildlife to mix. 

We also developed a machine learning model that looks for the same troubling patterns of deforestation in locations where outbreaks have never before been caused by spillover. This model was based on the same historic satellite images, as well as data about human populations in these areas.","We found one academic paper after another saying that deforestation is linked to outbreaks. But we also found that global health leaders aren’t acting on this information, preferring instead to focus on responding to epidemics already underway. The science shows that computational tools can help people take preventative measures to prevent an outbreak. So we decided to adapt the science to show readers exactly how deforestation might lead to the next big epidemic.

With powerful narratives, immersive photography, animated graphics and an unprecedented use of statistical modeling and satellite imagery, we translated complicated science into stories that explained why we need to act now, if we want to prevent the next pandemic.

Often, public health funding goes through a cycle of panic and neglect, with urgent calls for action during an emergency, then attention fades and people move on. Now is the time when post mortems of the COVID-19 pandemic are still underway, and governments are actively discussing how to spend funds to prevent the next one. Our hope is that through this explanatory series, we can help deepen the conversation about the root causes of outbreaks and help the public as well as policymakers explore new paths to prevention.",N/A,N/A,"All stories were in English. In addition, we translated the first story into French here: https://www.propublica.org/article/pandemie-saut-despece-epidemie-guinee-clairiere-de-foret",https://www.propublica.org/article/pandemic-spillover-outbreak-guinea-forest-clearing,https://projects.propublica.org/spillover/,https://www.propublica.org/article/how-propublica-used-machine-learning-investigate-where-ebola,https://www.propublica.org/article/pandemic-spillover-madagascar-health-in-harmony,,,,
"Rubber, Race, and Colonial Exploitation",Single project,Kontinentalist,Small,"Writer / Loh Pei Ying
Code / Tay Qijin
Data Design / Loh Pei Ying
Design / Munirah Mansoor
Editor / Gwyneth Cheng, Nabilah Said
Illustrations / Griselda Gabriele
Research / Heleena Panicker, Jasen Lo, Loh Pei Ying","Loh Pei Ying, Writer 

Pei Ying co-founded Kontinentalist in 2017. She leads the company in achieving its overall business and editorial goals, making strategic business development plans, and managing partnerships. Her work experience includes stints at various heritage-related institutions, including three years at the Asian Civilisations Museum’s curatorial department. Her background and passion for history is the driving force behind many of her stories at Kontinentalist, which delve into cultural and historical contexts. She seeks to empower voices through cause-driven stories as well as change people’s perception about Asia and its cultures.",Singapore,"Rubber, Race, and Colonial Exploitation explores the historical significance of the rubber industry in Singapore and its deep-rooted impact on the present discourse surrounding migrant workers in the country. The article highlights how European dominance in rubber supply, discriminatory land policies, rigged quotas, and labour systems have led to enduring inequalities, shaping the socio-economic landscape of modern Singapore. The legacy of such exploitation is traced through generational wealth disparities, particularly affecting the Malay population, and the continued reliance on migrant labour.","This story has become our top story on our website, and we think it’s because it showcases an element of Singapore history that is rarely talked about. We even received anecdotal evidence that it was used in the History exam paper for one of Singapore’s local secondary schools. 

Singapore’s and Malaya’s historical rubber trade is often written about in glorified ways, in terms of its role in enriching the country and building the generational wealth of many still-prominent families in Singapore and the region. Our angle of highlighting the discriminatory practices that existed, and connecting them to systemic injustices that continue to today bucks that prevailing perception and helps people think more critically about history and how it continues to impact us. 

The story received 7,475 views, with an average engagement time of 34 seconds. It also got 1,188 likes and 292 shares on Instagram.","As recent as July 2023, migrant workers continue to live and work in Singapore under a similar system that is deeply disadvantageous to them, with poor living conditions and employment rights. This article uses historical investigation to highlight how this system of exploitation is systemic and historic. It underscores the need for empathy, understanding history, and addressing systemic inequalities to mend the divisions in contemporary Singaporean society.

This historical investigation was laborious to produce. It first started with tedious research in the form of two decades of colonial office records in microfilms. Each annual report spanned more than a thousand pages, and it took a lot of time to sift through it to look for the relevant records. We retrieved data on wages in different regions, volume of migrants, the breakdown on their demographics were sourced from these detailed colonial office records. The story was further supported with supplementary research from journal articles, oral interviews, and further secondary data—this part spanned almost half a year. One of the key visual pieces was also the manual conversion of a historical print map of the spread of rubber plantations in Malaya into digital geospatial data that could be rendered into a map-scrolly.","Because the story was dense in historical fact, we needed a visually-driven and impactful narrative to bring the various components together. There were also multiple smaller stories, such as the different treatment of racial groups by colonisers that needed to be unpacked. To do so, we used large form, comic-style illustrations as the centre of the story, to bring empathy and life to a topic that was hard for laypersons to relate to. The visualisations were largely executed on Flourish and Mapbox, using a variety of dot plots, bar charts, treemaps, and scale comparisons to draw active comparisons across racial categories and business owners.","News isn’t always just about unfolding stories or contemporary events. Historical archives are rich in material, and it’s important to contextualise and look at structural factors when covering issues, especially those concerning minority rights. Historical material is also a great treasure trove in environments where data access and availability is poor, as in the case of Singapore.",N/A,N/A,English,https://kontinentalist.com/stories/rubber-race-and-colonial-exploitation,,,,,,,
Selling the Green Dream,Single project,Australian Broadcasting Corporation,Big,"Mark Doman, Pat McGrath, Katia Shatoba, Alex Palmer, Inga Ting, Kate Newton","This project was lead by Mark Doman and Pat McGrath.

Pat McGrath is a journalist with the ABC's investigations team.

Mark is a reporter and the visual journalism lead with the ABC's Digital Story Innovations team.",Australia,"In Australia, retirement savings contributions are compulsory and the funds that manage this money are required to publish where it’s being invested. Some of these funds market themselves as green or offer portfolios described as “sustainable” or “ethical”.

Our investigation focused on funds promising to limit investments in industries like fossil fuels, alcohol, or gambling.

We painstakingly compiled the disclosure spreadsheets from every sustainable-labelled retirement saving option in Australia. We then developed a methodology for checking how the fund’s claims stacked up against the reality of their investments. 

We found almost $1.2 billion invested in fossil fuel industry shares.","The story was read by hundreds of thousands of people, well above average for ABC News stories with strong engagement time.

Shortly after the story went live, one of the largest providers we investigated overhauled its website. They removed the slogan “Save the planet, by saving for your future” and the green marketing that had featured prominently was removed and replaced. At least one other superannuation provider, updated the fine print on its website to clarify its holdings. The financial regulator, ASIC, told us it would be pursuing any companies whose marketing did not match up with the fine print of their portfolios.

For the community, the story provided Australians with a tool that allowed them to easily assess and compare the share holdings of sustainable-labelled funds and to see how their retirement savings are being invested. This was one of the reasons we wanted to tackle this project, because for the average member of the community comparing the offerings of different funds is an onerous task, to say the least.","The wrangling of the data and coming up with a robust way of classifying, for example, a fossil fuel company were two big challenges.

There is no requirement for funds to publish disclosures in a consistent format. The spreadsheet structure changes between funds, assets are named inconsistently and columns, like unique identifiers for stock, are often missing. That meant we had to manually clean and copy over the data from each portfolio into our master spreadsheet. 

We then needed to develop a way to compare the shareholdings in the disclosures with a list of public companies globally that report revenue from fossil fuels, tobacco, gambling, alcohol, uranium and weapons manufacturing and military contracts. To do that, we needed a common identifier between the two datasets to join them. We first tried to match by stock name, but even after cleaning and simplifying the names, it didn’t produce robust results. Through our research, we came across the ISIN. A globally consistent, unique code for publicly listed companies. We sourced ISINs from public databases including stock exchanges around the world and matched them to our data. An arduous process that often required us manually researching companies to find the code.","Because there were usually 2 sometimes 3 people working with the spreadsheets at any given time, we kept all of our data in Google Sheets. We then hooked this into R, specifically R studio, for most of the data wrangling, cleaning and analysis.

We used chat-gpt to speed up the process of writing scripts for some tasks in R.

Because so much of the data entry was done manually, we used Tableau Prep in the final stages to spot check and interrogate our datasets for any inconsistencies against the original disclosures.

To visualise the data, the team used a combination of D3, Svelte and various in-house tools.","There are two key takeaways for me:

The first is the power of collaboration. The story combined the ABC’s data team with its investigations unit. Through this collaboration, we were able to not only interrogate the data but also dive into the reasons why this was an issue in the first place. It also afforded us more resources to track down people who had invested their retirement savings with some of these funds to get their reactions to our findings. The result meant we were able to produce an accompanying 10-minute video investigation as well that aired on the ABC’s 730 current affairs program.

The other is embracing the unknown. With this project, we knew the general approach we needed to take: if we could compile the disclosures and classify the public shareholdings by industry type, we could calculate how well super funds were living up to their promises. It sounded easy on paper, but in reality it sent us down the rabbit hole of security identifier codes and equity screening methodology. Through extensive research and speaking to numerous experts in the field, we were able to come up with a robust approach.",N/A,N/A,English,https://www.abc.net.au/news/2023-12-14/sustainable-ethical-super-funds-with-fossil-fuel-investment/103196032,https://www.youtube.com/watch?v=yBwE3JOMLWc,,,,,,
STAT Obesity Drug Tracker,Single project,STAT,Big,"J. Emory Parker, Elaine Chen, Allison DeAngelis","Emory Parker is the data project manager at STAT. In 2015, he was part of the team that won the Pulitzer Prize for Public Service for a series on South Carolina’s epidemic of criminal domestic violence. Prior to working in journalism, he worked in a neuroscience lab at the Medical University of South Carolina. Elaine Chen is a cardiovascular disease reporter at STAT, covering all aspects of heart and metabolic conditions including diabetes and obesity. Allison DeAngelis is a biotech reporter at STAT, covering biopharma startups and venture capital.",United States of America,"Novel drugs like Ozempic and Wegovy began a wholesale transformation of obesity treatment in 2023. In conjunction with our series covering this “Obesity Revolution,” STAT developed a unique data tool, the Obesity Drug Tracker, to provide an early look at even more powerful or targeted applications that companies hope to develop for consumers. The tracker is based upon meticulous analysis and curation of clinical trial data, analyst notes, press releases, and original reporting. It offers an up-to-date snapshot of dozens of new drugs under development in this rapidly expanding, highly competitive sector.","STAT readers were effusive in praise for the utility and innovation of this tool. One staffer in the U.S. Senate called the tool “head and shoulders above every other resource,” and said that this database would be instrumental in informing policymakers as they navigate this complex topic. In one month, the tool racked up over 15,000 visits and led directly to 33 new subscriptions to our premium service STAT+, with readers spending over 200 seconds on average on the page, all well above median values for typical posts on our site.","Building this tool required a great deal of reporting leg work as well as leveraging expertise from sources as well as STAT reporters and editors, in finding, interpreting, and parsing complex technical information from a wide array of sources. Few media outlets would be prepared to undertake such a reporting challenge. Meanwhile, it was a significant challenge to design the underlying database to be performant, flexible, and easy for non-technical reporters to use.","We designed a complex database with a custom editing interface powered by Airtable that allows reporters to quickly input and edit entries and cross-reference them with companies, drug categories, and development statuses. Node.js was used on the backend to power a microservice that pulls data out of Airtable and into a version-controlled repository that’s then mirrored on our CDN. A fully custom frontend was developed using the Svelte framework to both compute and visualize statistics about the data as well to allow readers to explore and filter the full dataset on their own.",Readers and the public can often be served by publishing versions of what would otherwise be only internal reporting materials. That collaboration between developers and reporters can turn such internal reporting into robust tools that are equally useful to reporters and to readers. Another lesson is that taking the time at the start of a project to properly design and implement a database with a robust schema and multiple tables will eventually yield a product that is more flexible and extendable than a simple spreadsheet.,N/A,N/A,English,https://www.statnews.com/2023/09/12/new-weight-loss-drug-tracker-novo-nordisk-eli-lilly/,,,,,,,
The Banned Books Project,Single project,The Marshall Project,Big,"REPORTING & WRITING: Andrew Rodriguez Calderón, Keri Blakinger, Liset Cruz, Cid Standifer, Vignesh Ramachandran, Shannon Heffernan
**COMMUNITY LISTENING:** Andrew Rodriguez Calderón, Ana Méndez, Vignesh Ramachandran, Nicole Funaro
**DATA PROCESSING:** Andrew Rodriguez Calderón, Cid Standifer, Aaron Williams
**VISUAL & MULTIMEDIA DESIGN:** Elan Kiderman Ullendorff, Bo-Won Keum, Celina Fang, Meredith Rizzo, Jovelle Tamayo
**VIDEO TUTORIAL:** Jasmyne Ricard
**STYLE & STANDARDS:** Ghazala Irshad, Akiba Solomon
**AUDIENCE ENGAGEMENT:** Ashley Dye, Chris Vasquez, Kristin Bausch
**DEVELOPMENT:** Ryan Murphy
**STORY, DATA AND VISUAL EDITING:** David Eads, Raghuram Vadarevu, Marlon A. Walker, Tom Meagher, Andrew Rodriguez Calderón","This project was conceived and executed by The Marshall Project staff across multiple teams..

Computational journalist [Andrew Rodriguez Calderón](https://www.themarshallproject.org/staff/andrew-r.-calderon) deserves special mention for his role in this project. As seen in the credits, Rodriguez Calderón contributed to the reporting, data processing, community listening, and visual editing of this package.",United States of America,"The furor around book bans sparked a critical national conversation. Often overlooked is the long history of censorship in prisons and jails. To address this, The Marshall Project investigated the challenges incarcerated people face in accessing and retaining books.

We published [reported essays](https://www.themarshallproject.org/2022/12/21/prison-banned-books-new-york-florida-censorship), a [searchable database](https://www.themarshallproject.org/2022/12/21/prison-banned-books-list-find-your-state), an AI-powered analysis of [state-level publication policies](https://www.themarshallproject.org/2023/03/16/prison-banned-books-policies-what-we-know), an [in-depth look at Ohio](https://www.themarshallproject.org/2023/06/20/ohio-odrc-prison-book-ban-java-hitler), an investigation into how [drug smuggling policies can turn into de facto book bans](https://www.themarshallproject.org/2023/10/18/prison-drugs-overdoses-book-banning), and a [reporting recipe](https://www.themarshallproject.org/2023/10/03/banned-books-prisons-how-to-report) so that other journalists can build on our work.","Among the project’s achievements:

* Turned the banning of books in prisons into a national issue and established The Marshall Project as a leading newsroom on the issue.

* Our data and reporting were central to a report from PENAmerica that kicked off the inauguration of their Prison Banned Books Week; thousands of articles were written about it.
 
* The American Library Association invited us to events to discuss our work with carceral librarians and other information scientists. They used our reporting and data to inform their upcoming Prisoner Bill of Rights with regards to access to information.
 
* The Texas Department of Criminal Justice rolled back their controversial drugs policy thanks to our story even before our story about book bans driven by drug smuggling concerns even published.
 
* Started a mailing list for The Banned Books Project that now includes around 70 people who have either expressed interest in our coverage or who work on the issue of banned books.
 
* Attracted industry attention for our approach to generative AI, which Andrew Rodriguez Calderón captured in write ups for industry publications, conferences, and ultimately helped spur the creation of The Marshall Project’s innovative AI Working Group.","We used collaborative design to place affected individuals at the center of our reporting and product. We employ [generative AI to distill bureaucratic text into concise summaries](https://generative-ai-newsroom.com/decoding-bureaucracy-5b0c1411171). The result is an innovative, evergreen package that examines the issue from the inside out, and demonstrates how to hold systems accountable while centering the information needs of people closest to those systems. 

The project represents significant innovation in making information accessible for the people most affected by the criminal justice system. As a package, this project expanded The Marshall Project’s audience more than any other project in 2023. 

Beyond our site, there were at least 40 stories in at least 30 publications, including the AP and numerous local publications. Reporters from The Marshall Project were interviewed for a multitude of radio and TV stories.

The investigative journalism of the (near) future will be persistent, multi-channel, and driven by engagement and data techniques that center audiences and meet information needs, empowering them to gather, verify and distribute local news and information. The Marshall Project’s coverage of prison censorship is a model for that future.","The data posed significant technical challenges on a state-by-state basis. There were many states with mixed values in different columns, some with no columns at all, just long strings of information, and others with information scattered across multiple sheets that had idiosyncratic formatting.

We also received many PDFs with lists that needed to be converted into text, extracted and transformed. In addition to the typical mix of data processing tools we often use to clean and consolidate data, we used AI to help us analyze state book ban policies.

We built an initial extract-transform-load pipeline ourselves using Python. We then contracted with Aaron Williams, an engineer at Netflix with a background in journalism, to improve our processing pipeline so that it’s easier to add and process new lists. 

ChatGPT helped us process and summarize state publication policies. Andrew Rodriguez Calderón wrote in-depth about our approach to “[decoding bureaucracy](https://generative-ai-newsroom.com/decoding-bureaucracy-5b0c1411171?gi=7336f1a5669e)” using AI for the Journalism AI blog.","We created a [reporting recipe](https://www.themarshallproject.org/2023/10/03/banned-books-prisons-how-to-report) that explains what other journalists can learn from this project. 

Anyone who wants to report on the data can learn…

* Why the issue of banned books in prisons matters.
* How to request book ban records from state prison systems, with sample requests.
* How to pitch stories and start reporting, with a pitch template.
* How to download and understand the data, with an explanatory video.
* Our editorial style and standards considerations about reporting on incarcerated people
* Other publications that have used the book ban data and reporting.
* How to embed the searchable table with a state’s banned books on a website.
* Social media language related to this project.",N/A,N/A,English,https://www.themarshallproject.org/tag/banned-books,https://www.themarshallproject.org/2022/12/21/prison-banned-books-list-find-your-state,https://www.themarshallproject.org/2023/10/03/banned-books-prisons-how-to-report,https://www.themarshallproject.org/2023/12/08/prisons-banned-books-censorship-2023,,,,
The Bat Lands,Single project,Reuters,Big,"Ryan McNeill, Deborah J. Nelson, Helen Reid, Allison Martell, Cooper Inveen, Matthew Green, Jake Spring, Ruma Paul, Andrew R.C. Marshall, Grant Smith, Michael Ovaska, Sreekanth Sivadasan, Ruma Paul, Andrew R.C. Marshall, Adolfo Arranz, Daisy Chung, Sam Hart, Jackie Gu, Simon Newman, Prasanta Kumar Dutta, Rupam Nair, Francis Kokoroko, Alphonso Toweh, Leonardo Benassatto, Bruno Kelly, Emma Jehle, Rafiqur Rahman, Sivaram V, Rosanna Philpott, Lucy Ha, Matthew Stock, Chen Lin, Matthew Weber, Janet Roberts, Feilding Cage, Paulo Prada and Blake Morrison",This project was reported and produced by a team of Reuters Journalists.,United States of America,"The Reuters project “The Bat Lands” relied on a groundbreaking statistical model and reporting on six continents to pinpoint where outbreaks of bat-borne viruses with epidemic potential are likeliest. The model, developed by Reuters reporters, analyzed natural and human-driven conditions surrounding previous outbreaks and identified where similar factors exist worldwide. Reporters mapped more than 9 million square kilometers – about 6% of the Earth and home to some 1.8 billion people – where conditions are ripe for a lethal bat virus to emerge. The human population in these areas grew by 57% over two decades, boosting chances of a new pandemic.","The Bat Lands analysis has proven its predictive power, and health experts have cited its value for preventing epidemics. 

One group involved in pandemic treaty discussions emailed the project to all negotiators, saying it shows the urgent need for global measures to prevent spillover – a pathogen’s leap from bats or other animals to humans.

Identifying areas with high spillover risk, like we did, helps companies and governments develop prevention and response strategies, two World Health Organization officials told us.

Our model has shown its predictive accuracy. Since 2020, the endpoint of our data, more than two dozen outbreaks have occurred in places we identified at highest risk. They include:
* [A deadly outbreak](https://www.reuters.com/investigates/special-report/global-pandemic-bats-prevention/) in October 2023 of the brain-swelling Nipah virus in a rapidly urbanizing region of India where our reporter dodged urine from bats flying overhead. 
* [Ghana’s first known outbreak of Marburg virus](https://www.reuters.com/investigates/special-report/global-pandemic-bats-jumpzones/), in 2022, near farmland littered with half-eaten fruit dropped by bats that can carry the pathogen. Farmers sometimes eat or feed the scraps to livestock. 
* [Fatal Nipah cases in early 2023 in Bangladesh](https://www.reuters.com/investigates/special-report/global-pandemic-bats-spillover/), where health workers with megaphones urged people not to consume date-palm sap. Bats contaminate vats that collect the syrup.","Aside from the complexity of the topic itself, requiring time-consuming reporting and data analysis, some significant hurdles turned this project into a multi-year undertaking. 

With 8 billion data points, computing power posed a challenge. Even with powerful servers, it took days to retrain the model and generate predictions each time we made changes.

Because we reported the project on six continents during the COVID-19 pandemic, illness and travel restrictions repeatedly interrupted our work. 
It took the Indian government four months to approve a journalist visa allowing us to report in Kerala for our [deep dive into the concept of spillover](https://www.reuters.com/investigates/special-report/global-pandemic-bats-spillover/).

It’s also notable that we were reporting in a crowded field. Other news organizations have published stories about the role of bats in pandemic risk. But none did a deep data analysis of that risk, a public service that offers the world a specific roadmap to areas that require stronger oversight and spillover prevention.

[ProPublica conducted its own machine-learning analysis](https://www.propublica.org/article/ebola-spillover-machine-learning-methodology-nigeria?utm_campaign=trueanthem&utm_medium=social&utm_source=twitter#:~:text=Random%20forests%20are%20a%20popular%20and%20versatile%20technique%20that%20has%20been%20used%20widely%20in%20academia%20and%20journalism) around the same time we did and included a link to our May series to support its own work published three months later. ProPublica’s stories were limited in focus to Ebola outbreaks in Africa.","We developed [our own methodology](https://www.reuters.com/investigates/special-report/global-pandemic-bats-methodology/), a groundbreaking statistical model that used an artificial intelligence tool to predict where bat-borne virus outbreaks might happen worldwide.

We began with data from Boston biotech Gingko Bioworks listing spillovers from 2002 through 2020. We did extensive reporting to verify and geolocate each case and added documented spillovers to expand the dataset.

For the same period, we built a spatial database, mostly from satellite imagery, of [56 environmental, social and economic conditions linked to spillover](https://www.reuters.com/investigates/special-report/assets/global-pandemic-bats-methodology/BatLandsCovariates.pdf). The 8 billion data points we amassed spanned the Earth and included temperature, precipitation, estimated number of bat species, deforestation, livestock estimates and urbanization.

Linking the 95 documented outbreaks to the conditions data created a profile of factors that existed where past spillovers occurred. We fed that information into a random forest model, a machine-learning tool, generating a score that told how closely each 25-square-kilometer section of the globe resembles areas with known outbreaks. The score predicts how prone each area is to spillover. We called the riskiest areas, those in the 95th percentile and above, “jump zones.”

This analysis formed the project’s backbone, showing us where to send reporters and driving graphics that illustrated the risk worldwide.","“The Bat Lands” is a monumental work of science journalism. Our reporters dove deeply into complex zoonotic and epidemiological concepts, reviewing hundreds of academic papers and interviewing more than 100 [scientists]https://www.reuters.com/investigates/special-report/assets/global-pandemic-bats-methodology/BatLandsScientists.pdf). They then conducted their own scientific inquiry.

The most important lessons, we think, are: Don’t shy away from original inquiry, and find experts to help. Review existing research and develop a deep understanding of your topic. Then find scientists and statisticians willing to help journalists explore new territory. We could not have done this project without their help.

Another key takeaway: Visuals, visuals, visuals. Our favorite response to The Bat Lands was: “I read your story. Well, more accurately, I experienced it.” That should be your goal. Think of visual elements as an integral part of the storytelling, and involve visual people from the start. We knew video, photos, maps and illustrations were critical to frame narrative spines for our stories and help readers digest the science and data. Before each reporting trip, we went over the goals of the mission, including the visuals we hoped to get on location. We also conceived and revised the visuals and text in tandem. This helped us tighten and focus the storytelling.",N/A,N/A,English,https://www.reuters.com/investigates/section/global-pandemic-bats-overview/,https://www.reuters.com/investigates/special-report/global-pandemic-bats-jumpzones/,https://www.reuters.com/investigates/special-report/global-pandemic-bats-spillover/,https://www.reuters.com/investigates/special-report/global-pandemic-bats-spread/,,,,
The Bruno and Dom Project: More than 800 million trees felled to meet Amazon beef demand,Single project,"The Bureau of Investigative Journalism, The Guardian, Le Monde",Big,"Andrew Wasley, Elisângela Mendonça, Youssr Youssef, Rob Soutar","The Bureau of Investigative Journalism is an independent, not-for-profit organisation that holds power to account. Founded in 2010 by David and Elaine Potter, we tackle big subjects through deep reporting that uncovers the truth. We tell the stories that matter.",United Kingdom,"To continue the work of the murdered Indigenous Peoples activist Bruno Pereira and British Journalist Dom Phillips, The Bureau of Investigative Journalism (TBIJ) produced new figures for Amazon deforestation linked to foreign beef demand. It was our contribution to Forbidden Stories’ Bruno & Dom Project, published on the anniversary of their deaths. 


We calculated that at least 800,000,000 million trees in the Amazon’s top beef producing states were lost between 2017-2022. Centering on abattoirs owned by Brazil’s ‘big three’ meatpackers that shipped internationally in the period, the investigation tangibly and originally linked Amazon deforestation to foreign consumption of beef.","The Guardian launched the Bruno and Dom project with a front page splash that trailed our beef investigation, raising the visibility of the specific threats to the Amazon and those reporting on them. As well as reaching hundreds of millions of readers across participating media, shortly afterwards and referencing the project, opposition MP Tanmanjeet Dhesi told the UK parliament: “it is important that while we protect our environment and the people, we also prevent those excellent environmental activists, journalists and indigenous activists who are fighting the good fight— not just on their own behalf, but on behalf of all of us—from coming to harm.” 


Members of the European Parliament, inducing those working on proposed supply chain due diligence legislation, were also alerted to our investigation and told us of its importance and leading environmental NGO Mighty Earth said it was a “Really impressive piece, that seems to have really resonated globally”. 


We also shared the data we commissioned with journalists in other countries, allowing them to link deforestation to their own domestic markets. Le Monde and Der Spiegel were among those to produce country-specific investigations based on our dataset.","President Jair Bolsonaro (2017-2022) cut funding for Brazil’s environmental protection agencies and publicly questioned satellite deforestation monitoring programmes. Under his presidency, which largely enjoyed the backing of agribusiness, forest loss and threats to Indigenous Amazon communities surged, as did the intimidation and the persecution of journalists, brought home chillingly by the murders of Bruno and Dom. But killing the journalist won’t kill the story, as is Forbidden Stories’ guiding statement. 


In this context, and with the government introducing restrictions on access to information that enables researchers to map Brazil’s beef supply chains, we shed new light on the specific threat of deforestation posed by cattle ranching – the number one driver of forest loss – in states both had spiked. 


Our research used widely accepted methods for calculating deforestation risk and was rigorous. But that didn’t stop the big meatpackers trying to delay and derail the story by questioning our methods.","We used official satellite monitoring programmes (Prodes and Deter) to calculate deforestation in the areas of top beef producing states that lie within the Amazon biome. We then plotted abattoirs owned by JBS, Marfrig and Minerva that were known to export internationally (using shipping data) to calculate the deforestation on farms within their “buying zones” the area (in square kilometres) within which the companies could reasonably be expected to source cattle, based on transport links and other factors. 


Comparisons of land area employed by journalists often rely on cities, countries or football pitches. So to bring our new firgure to life, and make the figure more tangible we used a method employed by Amazon research institute Imazon to calculate the number of trees that exist within a square kilometre in the biome – finding conservatively, that some 800,000,000 had been lost.","Collaborate. And if the data is hard to come by, commission your own.",N/A,N/A,"English, French",https://www.thebureauinvestigates.com/stories/2023-06-02/almost-a-billion-trees-felled-to-feed-appetite-for-brazilian-beef,https://www.theguardian.com/environment/2023/jun/02/more-than-800m-amazon-trees-felled-in-six-years-to-meet-beef-demand,https://www.lemonde.fr/planete/article/2023/06/02/de-l-amazonie-a-l-europe-la-filiere-du-b-uf-continue-d-exporter-de-la-viande-liee-a-la-deforestation_6175899_3244.html,,,,,
The food industry pays ‘influencer’ dietitians,Single project,"The Washington Post, The Examination",Big,"Caitlin Gilbert, Anahad O’Connor, Sasha Chavkin","Caitlin Gilbert is a Well+Being data reporter at The Washington Post. She earlier worked at the Financial Times and as a visual journalist at Reuters. She received her PhD in neuroscience and genomics from Rockefeller University.
Anahad O'Connor is a health columnist who writes about food and nutrition for The Washington Post's Well+Being desk. Before that he was a staff reporter for the New York Times, where he spent two decades covering health and science.
Sasha Chavkin is a correspondent for The Examination. He previously wrote for outlets including ProPublica and the International Consortium of Investigative Journalists (ICIJ).",United States of America,"An investigation by The Washington Post and The Examination found that the food and beverage industry has enlisted “influencer” dietitians to shape the eating habits of millions of social media users, encouraging them to eat more junk food and ignore public health nutrition advice. The influencers, many of whom are trusted registered dietitians, often failed to clearly disclose the industry groups or companies that were paying them.

We analyzed more than 6,000 pieces of social media content to determine the extent of industry influence, and found paid campaigns from the sugar and beverage industry trade groups and dietary supplement makers.","Two months after the story published, the Federal Trade Commission for the first time announced warnings against the two major food and beverage industry groups named in our initial story, and against the dozen nutrition influencers we named. 

The crackdown represented more aggressive enforcement of the FTC’s rules than had been previously enacted, and signaled that the agency was working to set a new precedent for holding both influencers and industry accountable for social media marketing campaigns. The action has the potential to dramatically change how influencers use social media; a review the reporting team did in December weeks after the FTC action found that everyone cited had removed older advertising posts — and had begun clearly marking paid and sponsored content. A lawyer in the e-commerce field told reporters that companies across all industries working with social media influencers were adapting their practices in light of the federal enforcement, and had started to update their contracts with influencers to detail proper disclosure standards.

A number of other news organizations — including CBS News, The Associated Press, NBC and Gizmodo — have followed with stories of their own based on the Post’s initial reporting and the FTC action.","Dealing with large quantities of social media video is always a challenge. Data reporter Caitlin Gilbert had to search across platforms to find users with specific phrases or words in their bios, find libraries that would download and transcribe large tranches of video, deal with metadata from two distinct sources, and had to find a way to access content such as Instagram Reels, which are substantially different from Instagram posts. Social media sites like TikTok and Instagram are built to prevent scraping, so Gilbert had to write scripts and use tools like puppeteer to circumvent rate limits and issues with lazy-loading content. At different points in the analysis, she used JavaScript, Python and R to approach specific technical challenges. 

Another component that made this story unique was the effort to ensure the investigation found it's audience -- the TikTok and Instagram users who were the targets of the industry’s influence campaign. The Post's Well + Being team published a TikTok explaining the investigation and its findings. That video went small-scale viral, generating over 100,000 views and 12,000 likes. When the FTC took action, the team created another TikTok about the crackdown. These were immensely important in driving traffic and conversation.","Caitlin Gilbert used puppeteer/node.js to grab the pertinent video links that met our standards, and then a bash script used the yt-dlp package to download the videos and faster-whisper, an AI-based package, to transcribe them. Analysis was done in R statistical software, and Google sheets were used for manual data entry/labeling to identify promoted content.
Because there was some concern that influencers would trying to avoid public scrutiny by pulling their videos from social media platforms when or before the story published, reporters also preemptively captured metadata and archived all video posts that the team directly reported on, both so that a record of the problematic content would be preserved but also to be able to compare old posts to newer versions, should the users edit posts or improve disclosures.","Working with social media video as data is technically tricky but can pay off with impactful stories that can shine light into largely unregulated corners of the world. In this case, we wanted to look at the intersection of the ever-growing world of ""influencer"" marketing and how it influenced our nutrition and eating choices. While nutrition influencers are a huge presence on social media, we purposely kept the scope of this project tight to increase accountability from both the industry and health profession standpoint.",N/A,N/A,English,https://wapo.st/3vER9MN,https://wapo.st/429Ow1K,https://wapo.st/3vER9MN,https://www.tiktok.com/@washingtonpostwellbeing/video/7278432035381824799,,,,
The Green Wave series: Examining the growing influence of Islamist-conservative politics using electoral data,Single project,Malaysiakini,Big,"Andrew Ong, Aidila Razak, Ooi Choon Nam, Hariz Mohd, Emylia Rosli, Pia Ernadya, Raveena Nagotra, Syariman Badrulzaman, Noor Rahmat Shah Haron, Mahen Manickam, Lee Weng Keat and Tan Hong Kai.","The Kini News Lab is a team specialising in data journalism and visual storytelling. It is part of the Malaysiakini newsroom. Malaysiakini is an independent news organisation and is Malaysia’s most read and oldest news website. The Kini News Lab was founded in 2019 as a way for Malaysiakini to experiment with innovative ways of doing journalism. 

It is a small team of four people, but works collaboratively with members of the wider Malaysiakini newsroom to produce groundbreaking work, lifting the standards of Malaysian journalism.",Malaysia,"Through detailed analysis of electoral data and in depth interviews with stakeholders, this two-part project uses the detailed results of Malaysia’s 15th general election to examine the growing influence of an Islamist-conservative political coalition at the polls. 

It not only dissects the election results down to each polling station, but also allows users to analyse the results of their own local areas, review the data by voter demographics and predict the outcome of the upcoming state elections, giving them access to the data at a level never before possible.","The series generated more than 240,000 views. They also have high engagement rates, with readers spending about five minutes on the page.
 
Two forums in two states were held to discuss the projects and it attracted a full house audience of about 100 participants each, not including live stream audience. We were also invited to speak on multiple TV and radio programmes following the publication, or contacted by other newsrooms to collaborate for future elections.

In our view, the biggest impact has been the democratisation of electoral data. The data cost almost double the median salary in Malaysia, and for the first time, the public could access it for free (in Bahasa Malaysia) or for a small fee of RM60 (USD12.90) for three months access (in English and Chinese). Since publication, many have shared their own analysis of electoral data on social media, using our project.

In a media interview, an electoral watchdog cited the project and said it showed the value of each vote. He noted that since many seats were marginally won, our widget showed how election results can change completely if as little as 10% of voters stayed home.","While it is public data, the only way to access the election results down to each polling station is by purchasing the data from the Election Commission. The Kini News Lab paid RM4,426 (US$951) for the dataset. This is close to double the median wage of formal workers in Malaysia, and is therefore inaccessible to most Malaysians and Malaysian newsrooms. 

Despite the hefty price, the data did not come in a form which is easily used for analysis. As such, we had to spend more time and funds converting the data. 

We were also operating without a UI/UX designer, so faced the challenge of converting the complex data into something easily used by people who are not ‘election nerds’.

The investment was, however, worthwhile for Malaysiakini because the analysis would provide tremendous insight into what might be the outcome of the then-upcoming six state elections. 

Further, the insights were important because the 15th general election resulted in a hung parliament, and the subsequent government was formed through political negotiations. Understanding the likely trend which led to this, and seeing if this trend is likely to persist, gives deeper understanding into the political instability which lies ahead.","We utilise Adobe Acrobat Pro to convert more than 2200 pages of PDF files to Excel as Tabula, Microsoft Excel, and Camelot led to duplicated entries. We also used Python to clean and convert the data.

For building the webpage, we employ React Nextjs and the Tailwind framework, along with various libraries like Mapbox, Highcharts, and react-scrollama.

We used JavaScript to develop a simulation calculator enabling users to predict election results by manipulating turnout and voting preferences.

A challenge we faced involved dealing with a large map file. To address this, we simplified the geojson file using mapshaper. To ensure smooth loading, especially on mobile devices, we merged all election data with the geojson, compressed it using tippecanoe, converted it into a tileset, and uploaded it to Mapbox.

Additionally, we integrated a temporary geocoding API from Mapbox, allowing users to search for their location and access election results at the polling district level.","One of the biggest obstacles of procuring data in Malaysia is the cost. This project taught us that for some projects, the investment is worthwhile.

As we had made a significant investment to purchase and analyse the data, we felt the urgent need to make sure the project has a greater reach outside the traditional Malaysiakini audience. As such, we invested in organising public forums, as well as allocated time to do media interviews on radio and television. We also leveraged on our networks to pitch to other media, to alert them of what we’re doing and that we are open to being interviewed on air about our work. 

As we did not have additional funding to ‘market’ the project, we also started curating a Whatsapp mailing list and recruited the help of the entire newsroom to share the link to influential individuals like politicians and civil society members, who we feel would be interested in this topic, and would likely share with their peers. As a result, we managed to recoup the expenses paid through new subscriptions, and grew our audience base. 

Since this project, we have tried to employ all these learnings on subsequent projects to considerable success.",N/A,N/A,"Bahasa Malaysia, English, Chinese",https://newslab.malaysiakini.com/greenwave-state-polls/en/,https://newslab.malaysiakini.com/greenwave-generational-gap/en,https://www.bfm.my/podcast/evening-edition/inside-story/understanding-state-election-predictions-and-why-every-vote-counts,https://en.syok.my/podcast/episode/103738,,,,
The Memorial of Calais,Single project,Wedodata and Les Jours,Small,"Maël Galisson, Elodie Cassignol, François Prosper, Karen Bastien, Clément Thorez","Maël Galisson, journalist specialized in stories of exiles and borders 
Elodie Cassignol, web project manager
François Prosper, datadesigner
Karen Bastien, datajournalist
Clément Thorez, datadesigner",France,"Since 1999 and until 2023, 367 migrants have died between the Franco-Belgian border area and the United Kingdom. This is what ""Les Jours"" counted in a work as unique as it is exceptional. Name, first name, age, nationality, migratory route, circumstances of death, photos... For several years, we searched and compiled all possible information on the exiles who disappeared along this maritime border. Our “Calais Memorial”, to be discovered below, lists the victims so that they are not forgotten. Interactive datavisualisation, infographics and video mapping: 3 different formats have been created for this unique project.","A lot of reactions on the political level because as is often the case, what is not counted does not count! So being able to document these deaths precisely and over a long period of time was a shock in France. It was also recognition of the field work carried out every day by NGOs in this particular area around Calais.","This is a unique project because no one has ever taken the time in France to create a precise database on these men and women who died on the Franco-British border. Beyond nationality, ages and sexes give indications on the types of populations pushed into exile. But above all, the location and context of their death clearly illustrates the fortress that Calais has become, pushing these refugees to take greater and greater risks. By datavisualizing these lives, we also illustrate a policy of closure that often goes unnoticed.","We choose to offer 3 different types of visualizations and therefore we relied on several technologies:
- interactive directory of deceased people in HTML/CSS
- summary dataviz of all the people documented in Illustrator
- animated video mapping with CineDataMap, a video mapping tool created in-house","In the era of Big data and open data which puts millions of pieces of data within reach of editorial staff, it is important to remember that the manual work of investigation, the collection of information person by person, certificate of death by death certificate, interview by interview still remains very valuable for monitoring, reporting and giving substance to phenomena that political powers do not want to see, or even want to hide.",N/A,N/A,french,https://lesjours.fr/obsessions/calais-migrants-morts/ep1-memorial/,https://wedodata.fr/productions/lesjours-morts-calais#,,,,,,
The Missiles on Our Land,Single project,"Princeton Program on Science and Global Security, Columbia School of Journalism, Nuclear Princeton, Scientific American",Big,"Sébastien Philippe, Ivan Stepanov, Svitlana Lavrenciuc, Katie Watson, Nina Berman, Ryo Morimoto, Zia Mian, Jeffery DelViscio, Duy Linh Tu, Sebastian Tuinder, Dominic Smith, Joshua Pribanic, Joseph Polidoro, Ella Weber, Tulika Bose, Todd Whitney, Jessica Lambert, Blue Carlsson, Lilian Fitzgerald, Joshua Wor","This project is a collaboration between Princeton University's [Program on Science and Global Security](https://sgs.princeton.edu/), [Columbia School of Journalism](https://journalism.columbia.edu/), [Nuclear Princeton](https://nuclearprinceton.princeton.edu/), and [Scientific American](https://www.scientificamerican.com/).

Principal Investigators: Sébastien Philippe (PI, Princeton University), Nina Berman (co-PI, Columbia University), Ryo Morimoto (co-PI, Princeton University), Zia Mian (co-PI, Princeton University)

Lead editor: Jeffery DelViscio (Scientific American)

Data: Sébastien Philippe (Production, Scientific Research, Data Analysis), Ivan Stepanov (Software Engineering, Data Analysis), Svitlana Lavrenciuc (Digital Cartography, Data Visualization), Katie Watson (Design, Production | Brown Institute for Media Innovation)

Additional team credit available at: https://missilesonourland.org/about.html",United States of America,"The Missiles on Our Land investigates the human and environmental risks associated with the most significant nuclear weapon build-up in the United States since the end of the Cold War. It is the result of collaboration between Princeton University's Program on Science & Global Security, Nuclear Princeton, and Columbia University’s School of Journalism, published in partnership with Scientific American. The project combines large-scale computational simulations of the consequences of nuclear war with ethnography and journalism to explore and shed light on the remaking of the US’s nuclear arsenal from multiple scientific and social angles.","The project maps the consequences of a concerted nuclear attack on US land-based nuclear missiles that are meant to serve as targets in a nuclear war. We found that nearly the entire population of the contiguous US and the most populated areas of Canada, as well as the northern states of Mexico, could be at risk of lethal fallout, depending on the weather conditions at the time of an attack. We identified communities most at risk, including the Three Affiliated Native American tribes of North Dakota – the only tribes hosting nuclear weapons in the United States, and investigated how the US Air Force was presenting its project to deploy new nuclear missiles on their reservation for the next 60 years.

The project was well received by the public, academics, and at-risk communities, including members of the Three Affiliated Tribes. The authors have been presenting their findings at leading universities, think tanks, the United Nations, and on Capitol Hill to members of Congress as well as on the Fort Berthold Reservation. The data was released in an interactive format that other journalists, researchers, and members of the public can use and the project has already been included in course syllabi.","Five US states, Colorado, Wyoming, Nebraska, Montana and North Dakota, are home to 450 nuclear armed land-based ballistic missiles, kept on continuous launch alert and intended to deter nuclear attack by a foreign adversary. Over time, it has come to be understood that these sites would be a “sponge” for a foreign attack, as an adversary would need to focus their efforts on hitting all these sites simultaneously to avoid a retaliatory strike. As part of current $1.5 trillion plans to modernize the US nuclear arsenal, existing missiles at these sites are intended to be replaced with new versions and infrastructure refurbished. However environmental impact reports have not assessed the risks associated with an actual attack.

We reviewed thousands of pages of Air Force environmental impact documents. We conducted dozens of interviews in Washington DC, North Dakota, Montana, Wyoming, Colorado, Nebraska, and New Mexico with nuclear weapon experts, military officials, and members of the public, including those living amid the nuclear missile silo fields. We performed hundreds of fallout simulations to show the risk to local populations as well as people across North America, including predicted fatalities and radiation exposure, to a level of detail never previously possible.","A concerted nuclear attack on the missile silos in the U.S. heartland would generate radioactive dust that travels with prevailing winds. We used high spatial and temporal resolution archived weather data to simulate the paths of the resulting plumes for 48 hours, by when most of the dust settles. Because wind directions change daily, we computed fallout dispersal from an 800-kiloton warhead detonating simultaneously at each of 450 silos on any given day of 2021. This required developing custom code to create input files for the US NOAA Hysplit atmospheric transport software and post-process terabytes of output data. To accommodate the unprecedented size of the grid covering North America at a 1-km resolution, simulations were run in parallel on 128 cores with 1 terabyte of RAM. Our results map for the first time the areas most at risk of fallout and the worst case weather-dependent scenarios for the United States, Canada and the northern states of Mexico. We then combined our fallout data with 1-km resolution population and building height density to calculate the resulting fatalities accounting for sheltering. Depending on wind directions, a nuclear attack on the missile silos could kill several million people.","This project brought together scientists, journalists, scholars, and editors at Scientific American to study and report on the environmental and societal risks associated with the largest US nuclear military program since the Cold War. It took two years from beginning to publication. More than twenty people worked together on producing the data at the core of the project and the associated multi-media package, including a webstory, a documentary short, an interactive map, a five-part investigative podcast centered on Native American communities, and two additional long form reporters and half a dozen editors also contributed to a special report by Scientific American (December 2023 issue).

This project benefited from a combination of strong editorial support at the onset, built-in scientific expertise, large computational resources, Native American researchers and reporters, experienced photojournalists and filmmakers, media innovation scholars, as well as data scientists and visualization experts. The biggest challenge was the project timing and management involved in releasing all the assets simultaneously. 

The result demonstrates how deliberately combining multiple scholarly disciplines with journalism practice from the beginning of a project and until its completion can result in innovative reporting on military sensitive projects in the public interest.",N/A,N/A,English,https://missilesonourland.org/,https://www.scientificamerican.com/report/the-new-nuclear-age1/,https://fallout.brown.columbia.edu/,https://www.youtube.com/watch?v=Uuf7uLujKDs,,,,
The Odin Project,Single project,Cuestión Pública,Small,"Claudia Báez, Esteban Ponce de León, Diego Páramo,Sergio Rojas, Jorge Posada, Nelson Casallas y Carlos Rodríguez.","Claudia Báez, an award-winning investigative data journalist, is the CEO and co-founder of Cuestión Pública. Esteban Ponce de León, a Mexican data scientist, develops AI-based open-source tools. Diego Páramo, is a renowned international speaker, visionary CEO, and serial founder. Sergio Rojas, a senior IT consultant, leads groundbreaking in Generative AI. Jorge Posada, a Data Architect, excels in Natural Language Processing and Data Science. Nelson Casallas is an acclaimed award-winning designer, and Carlos Rodríguez is a versatile video producer and entrepreneur. Part of this team includes winners of the 2023 Sigma Awards.",Colombia,"Odin links investigative journalism with social media trends to provide automated content that brings context to the Cuestión Pública (CP) audience. Investigative media outlets must stay relevant to the media agenda while undertaking their investigative pipeline. Covering current events interrupts proper investigative journalism (IJ) in concentration and time. Over five years, CP through two comprehensive data-driven investigations (Game of Votes, We Know What You Did Last Legislative Period) compiled 4,230 entities —companies, individuals—. Odin connects current events with our databases to generate happening content enriched with IJ and enhances the workflow efficiency of the team using LLMS and AI technology.","Odin, an ambitious platform developed at the AI in Journalism Challenge (AIJC), stood out as a top 5 finalist, selected from 12 newsrooms among 113 global applicants. Cuestión Pública, the team behind Odin, earned an honorable mention at last year's Splice Beta Journalism Festival in Thailand.

The Odin Project utilizes large language models and generative artificial intelligence technology to enhance the creation of current events content enriched with investigative journalism context. This initiative is a unique approach to ""zero waste"" in investigative journalism, aiming to make the most out of the available information and data. The customers of Odin are the investigative reporters of Cuestión Pública. 

In Odin's first implementation stage target in 'X, ' the system reduced the production time of CP threads in our newsroom's X social media from two hours to 15 minutes. Cuestión Pública journalists saved time on investigative reporting while improving workflow (over 100 threads sample). CP engagement increased 4% during H2 2023 thanks to improved coverage of current events. We're refining Odin's tone and voice on ‘X’ and adjusting the integration system with our database. In the next stage, we aim to transform this content for multi-platform use.","CP, specializing in investigative journalism, takes months for in-depth investigations. Over four years, it's broadened coverage to include breaking news and current events linked to these investigations. This strategy has improved audience insights into key political figures, merging timely news with thorough reporting, leading to audience growth, media recognition, and enhanced brand value.

In 2023, the outlet intensified its social media activity, particularly in covering current events, leading to greater versatility among its investigative journalists. However, this shift caused longer investigation times due to diminished focus. To address this, CP developed the Odin interface to streamline journalists' coverage of current events.

Collaborating with Pulse AI, Cuestión Pública faced technological hurdles in developing Odin. A significant challenge was organizing and cleaning data (4.230 entities) related to the news apps ""Game of Votes"" and ""We Know What You Did Last Legislative Period,"" spread across over 85 Google spreadsheets. The task involved unifying disparate entities (companies and individuals) and identifying duplicates, leading to potential gaps in Odin's current results.

We faced challenges in adjusting Cuestión Pública's tone and voice in 'X' threads created by Odin. Initially using few-shot learning, we're shifting to fine-tuning to improve results.","Through streamlined technology, Odin aims to enhance Cuestion Publica's productivity and media visibility.

We used the few-shot learning method to train Odin's AI platform. We connected to the Open AI API and used the generative AI model ChatGPT to do this. Additionally, we employed transformer architecture to process the text from our databases of journalistic investigation facts, fragmenting it to obtain a vectorized database (using embeddings).

We used ElasticSearch technology to store this data. We employed the RAG (retrieval augmented generation) method to extract relevant information.

For the next version, we are implementing a fine-tuning method to improve Odin's AI in generating content tailored to the voice and tone of Cuestión Pública in the final output accuracy.

Our technical approach has allowed us to develop an efficient and accurate AI platform. David Caswell and Bahareh Heravi supported Cuestión Pública and the Pulse AI team during weekly meetings to advance the Odin AI model for the AIJC program.","Through the approach of 'zero waste' in investigative journalism, the Odin project aims to help independent media focused on investigation leverage their deeply researched content previously published over the years and their databases consolidated through investigative efforts to connect it with current events in their countries. This approach will increase their team's productivity and its media visibility. 

Investigative journalists no longer have to wait months to publish their work and gain public attention. Connecting previous data and information to current events can help investigative media outlets stay relevant and improve their presence on the public agenda. Odin automates the process of investigative journalism teams of these outlets, optimizing their time and eliminating manual work.
 
Odin plans to create multi-format content (video, audiogram, IG's carousels, comics) on various social media platforms in the next version, with increased resources and development. Also, the Odin Project aims to identify trending topics by using social media APIs or listening tools to flag people and companies in the Cuestión Pública database and suggest content. Currently, users are",N/A,N/A,Spanish,https://youtu.be/A048gmP6Xec,https://youtu.be/gVVlO8ovvLk,https://cuestionpublica.com/mencion-de-honor-para-cuestion-publica-en-la-competencia-global-de-inteligencia-artificial-en-periodismo/,https://x.com/cuestion_p/status/1737251364663697793?s=20,,,,
"The stolen treasures. The 110,000 artifacts from Ukraine found in two Russian museums.",Single project,Texty.org.ua,Small,"Inna Gadzynska: idea, analytics, text;
Nadia Romanenko: data collection and processing;
Nadia Kelm: design, graphics;
Yevheniia Drozdova: layout, interactive.","**Inna Gadzynska** experienced journalist and editor; the deputy managing editor in TEXTY since 2020. Also worked as a media-trainer and media-expert for several international institutions.
**Nadja Kelm:** designer at Texty.org.ua. Author of the course about design in data visualization for School of Infographic (Internews Ukraine), winner of the national competition The best book design 2019 (the Arsenal Book Festival).
**Nadiia Romanenko**: is an analyst and data journalist at Texty.org.ua, where she started as an intern in 2015. Her scope of interest is studying manipulative news and Russian disinformation in Ukrainian online media using natural language processing and deep learning.",Ukraine,"In this study, we talked about **110,000 archaeological pieces of Ukrainian origin, which are kept in two Russian museums** - the State Hermitage Museum in St. Petersburg and the Russian Historical Museum in Moscow.
Icons, jewels, weapons, books, valuable artifacts — all of this was taken out of Ukraine by the Russians for centuries. By taking these artifacts from Ukraine, the Russians tried not only to demonstrate their connection with Kievan Rus, but also their deep historical roots. They continue to do this recently, by taking away valuables from museums in the occupied Ukrainian territories.","Our investigation got wide publicity, first of all within Ukraine, but worldwide too. In Ukraine we detected reprints and publications in at least 20 other media outlets, and up to 10 from abroad.
The project was granted to a[ TOP-10](https://gijn.org/stories/editors-pick-top-10-data-journalism-projects-from-2023/) data-journalism projects of the 2023 by GIJN.
""The Ukrainians"" project published [an interview ](https://theukrainians.org/povernuty-vse/) with the author on how this project was born.
""Ukraine: The latest"" podcast by The Telegraph also broadcasted [a talk with Inna Gadzynska](https://podcasts.apple.com/gb/podcast/moscow-ramps-up-attacks-how-ukraines-treasures-ended/id1612424182?i=1000634793854) about the project. 
21,000 views on website; and about 125,000 in social media.","In October 2023, a month later, National Agency for Preventing Corruption of Ukraine (NAPK) launched new component ""Stolen Heritage"" within its web project Art&Sanctions: https://sanctions.nazk.gov.ua/en/art/stolen/ . _The values stolen by the Russians are used for the propaganda purposes of the Russian Federation, in particular, to fuel the ""Russian world"", demonstrating the so-called ""rich Russian culture""_, the NAPK emphasizes. This fact confirms the high-level of relevance of the TEXTY project ""The stolen treasures"", which gives clear view on the scale of the problem.","The research was complex and lengthy - first we wrote a special algorithm that scraped all the information about the online exhibits of the Hermitage and the Historical Museum of the Russian Federation. Next, our journalists and data researchers manually ""cleaned"" this data to select only those that were moved from the territory of Ukraine. For the most part, finding museum pieces originating from the territory of the modern Ukraine was riddled with difficulties. Firstly, the majority of exhibits have missing data under the headers “Place of Discovery”, “Excavation Site”, “Country of Origin” etc. Although the “Country of Origin” information may simply say “Byzantium”, “the Roman Republic”, “Bosporan Kingdom'' or “Eastern Europe”, this may as well mean that the treasures were brought from Crimea and other parts of the modern Ukraine. Russians tend to name Ukrainian locations under Russian Empire titles. Since the locations could not be filtered automatically, it had to be done manually. To show where exactly in Ukraine the treasures and artifacts were found and where they were taken from, we added approximate geodata to each find.","_Inna Gadzynska_, the author, read about many jewels moved from Ukraine in different historical periods in the monograph of Ukrainian historian Serhiy Kot, which she found on the Internet in the late fall of 2021. In the 900-page work on restitution problems, it was described in detail how Ukraine lost its cultural values in different historical periods and what role Russia played in this. In february 2022 Russia started full-scale invasion in Ukraine and Inna returned to the topic of stolen cultural values only in April 2023 and had a plan to talk with Serhiy Kot. However, it turned out that there is no one to write or call. Serhiy Ivanovich Kot died suddenly in March 2022. 
But TEXTY didn't give up and tried to find and understand something on their own by researching Russian museum funds. Actually, the story about 110,000 arifacts in two Russian museums is an attempt at such a systematization.",N/A,N/A,"Ukrainian, English",https://texty.org.ua/d/2023/stolen_heritage/en/,https://texty.org.ua/d/2023/stolen_heritage/,https://podcasts.apple.com/gb/podcast/moscow-ramps-up-attacks-how-ukraines-treasures-ended/id1612424182?i=1000634793854,,,,,
The year of the Shahed,Single project,"Airwars, Der Spiegel, Financial Times",Small,"Sanjana Varghese, Nikolaj Houmann Mortensen, Julia Nueno, Azul de Monte, Joe Dyke, Rowena de Silva (Airwars)
Oliver Imhof, Alexander Epp, Niklas Marienhagen, Chris Kurt (Der Spiegel)
Chris Cook, Max Seddon, Anastasia Stognei, Felicia Schwartz (Financial Times)","Airwars is a civilian harm monitoring organisation, based in London. The Airwars Investigations Unit uses a range of cutting edge journalistic methods to tell compelling stories about conflict, in partnership with leading newsrooms around the world. For this investigation, Airwars partnered with Der Spiegel on a range of stories, and the Financial Time on one story, around the proliferation of the Shahed.",United Kingdom,"In the first year of their usage in Ukraine, nearly 2,000 Shahed drones were launched into Ukrainian territory by Russian forces, according to data-focussed investigation by Airwars. Using open source, traditional and visual journalistic methods, it showed how Russia turned a cheap Iranian suicide drone into a vital component of its war.

By tracking every reported Shahed launch over 12 months, Airwars' investigation - in partnership with Der Spiegel - illustrated that while the Shahed first emerged as part of the Russian arsenal a year ago, the tempo and intensity of attacks across Ukraine escalated significantly during 2023.","The impact of this article was significant in the short term, and we believe that there are ongoing projects which this investigation will contribute towards. This article reached a wide audience online – the accompanying publication with Der Spiegel was the most read on their website for that entire weekend. It was retweeted and shared widely on Twitter / X, with several prominent drone experts and commentators praising its findings. It was also re-published, in a different format, with Dagbladet Information, the daily Danish publication. 

Sanjana Varghese presented the findings at the European Center for Constitutional and Human Rights on a panel discussing the development of a new human right: safety from above. We have also worked with a group of lawyers seeking to pursue various civil and criminal litigation routes to hold Iranian manufacturers to account for their role in the war in Ukraine. This has involved providing the underlying data and trends analysis for our public reporting to lawyers intending to file a case in the United States in early 2025. A number of academics have also reached out to gain access to the data for their research.",N/A,"The investigation utilised open source tools, such as geolocation on Google Earth Pro and flight and ship tracking, to find and corroborate information. Through producing a unique dataset, we developed a unique way to measure the impact of the Shaheds, which had attracted significant media attention since it was first used due to its Iranian origin. We also corroborated this with native-language reporting and incorporated audio into the article, centering civilians' personal experiences of how the Shaheds impact daily life.

We also used specially-created animations, as well as satellite imagery, to demonstrate where these Shaheds were being launched from and where their component parts originated. We monitored flight tracking and ship tracking software to correlate potential weapons deliveries with dates of launches / usage, as a method of corroboration. Through a combination of open-source reporting, interviews with Ukrainians, trade data and in-depth analysis, we produced an original investigation with a human element.","We believe that there are several methods used in this project which could be replicated to great effect in other newsrooms. This article used innovative, open source reporting methods to investigate the pervasive use of a particular weapon to terrorise Ukrainian civilians. The format of the article, as an immersive multi-media feature, brought readers into understanding this now ubiquitous weapon in a completely novel, experiential fashion. For example, we used information gathered via open source methods to produce a novel dataset, taking a step back from daily news coverage of the Shahed to provide a larger picture. We also spoke to citizens on the ground in Ukraine using voice notes, and incorporated their actual voices into the article, rather than just as text. This helped ensure that a data-focussed story felt human and had the desired powerful effect on the reader.
This investigation helped us demonstrate how to marry traditional reporting skills, such as incorporating human and expert voices, with open source intelligence, to demonstrate how innovation around these tools within a newsroom can actually enhance traditional journalistic practices, not overpower them.",N/A,N/A,"English, German,",https://airwars.org/investigation/shahed-year-russia-ukraine-iran/,https://www.spiegel.de/ausland/russland-setzt-auf-iranische-kamikazedrohnen-im-ukraine-krieg-ein-jahr-drohnenterror-a-05333ab2-4705-407f-8a4e-10b7f12ec257,https://airwars.org/investigations/the-new-russian-drone-linked-to-iran-backed-factory/,,,,,
Tracking conflict from space,Single project,The Economist,Big,"Ainslie Johnstone, Sondre Solstad, Andreas Moor, Filippo Teoldi, Rosamund Pearce.","Ainslie Johnstone is a data journalist based in London. She joined the newspaper in 2021.

Sondre Ulvund Solstad is _The Economist_’s Senior data journalist. Sondre started at _The Economist_ in February 2020. 

Filippo Teoldi is a visual interactive journalist based in London. He joined the newspaper in 2023. 

Andreas Moor is a visual interactive journalist based in Switzerland. He joined_ The Economist_ in 2022. 

Rosamund Pearce is a visual journalist based in London. She joined in 2019.",United Kingdom,"We used satellite imagery and modelling to gain new insights into the conflicts in Ukraine, Sudan, and Israel/Gaza; tensions around Taiwan; as well as the aftermath of natural disasters. To do so we developed two techniques based on open-source satellite data. These allow us, and others who use our tools, to track the location and intensity of fighting and the damage it causes: mapping war in near real-time, and its destruction building-by-building. This enabled us to cover conflict in an entirely new way, and also benefited others in the open source community, academia, and humanitarian organisations on the ground.","These projects have been key elements of our reporting in some of the year's biggest stories and have updated continuously since they were launched: our Ukraine “war-fire” tracker since June and Gaza building damage tracker since October. 

Academics in Germany, Norway, Italy and the United States use our “war-fire” models to track conflict, some applying the models we have developed to new conflicts (e.g. in Somalia). In Sudan, local refugee-aid groups have used our maps of likely fighting to direct efforts to provide mental health aid. At least one British NGO—Insulate Ukraine— uses these maps to direct aid. 

Our map of building damage in Gaza has served as a cornerstone of our coverage of the conflict, and on October 19th, we provided the first quantification of buildings damaged in Gaza, a metric used by most major news publications covering the conflict. Both tools have been cited in academic articles.","We would like to gratefully acknowledge the way in which our methods draw on the work of the world-wide OSINT community. For instance, many there have used thermal anomalies to investigate individual conflict events-from which we draw inspiration for our automated system covering entire countries. The UN has previously published work on how to use SAR to track urban damage after natural disasters which provided the framework for our analysis of wartime damage to cities. We also would like to acknowledge help from academics who we have consulted on these projects, including Daniel Arribas-Bel and Francisco Rowe at the University of Liverpool and Ollie Ballinger at University College London. 

While this entry highlights our efforts to cover conflicts in Ukraine, Israel/Gaza and Sudan, we have used similar or other satellite methods to cover tensions around Taiwan, as well as the aftermath of natural disasters in Turkey and North America.","Our fire data comes from NASA’s FIRMS system, a satellite-based system set up to detect wildfires. This detects temperature anomalies, whether they are caused by wildfires, war, farming or industrial processes. We trained an ensemble of machine learning models to automatically mark which of the fires FIRMS detects are likely to be related to military activity. We then use this system to automatically categorise tens of thousands of fire events (updating daily). We have before and since verified that these “war fires” line up well with other reporting on the state of conflicts over time, where available. 

Our building-damage analysis is based on synthetic-aperture radar imagery collected by the ESA Sentinel-1 satellite. We used a pixel-wise t-test to compare images collected after the war or natural disaster event, with those taken before. We then merge this map of signal change with building footprints data from Microsoft Buildings or OpenStreetMap to calculate a probability that each building has been damaged. In Ukraine and Gaza we have calibrated this to visual surveys of damage in conflict zones conducted by the UN. Our infrastructure-damage analysis pipeline was developed with the help of Ollie Ballinger at UCL.","When a war breaks out, or a natural disaster strikes, collecting accurate data is hard. The few numbers coming out of conflicts or disaster zones are often difficult to verify and quick to change. However, such data is crucial: both to understand and respond to conflicts and their aftermath. Satellites offer significant advantages in covering wars: the data they present is regular and consistent, and can provide visibility over an entire conflict zone—all things that are very difficult to achieve when reporting from the ground. They cannot themselves provide a comprehensive view of the conflict but can provide a useful supplement to other, more traditional, reporting techniques (as well as to other open-source intelligence methods). To help other journalists, we have made our data and methods open source where possible. We hope this will enable more organisations to use open-source satellite data to cover conflict better.",N/A,N/A,English,https://www.economist.com/interactive/briefing/2023/02/23/data-from-satellites-reveal-the-vast-extent-of-fighting-in-ukraine,https://www.economist.com/interactive/graphic-detail/ukraine-fires,https://www.economist.com/graphic-detail/2023/07/20/data-from-satellites-suggest-violence-has-surged-in-much-of-sudan,https://www.economist.com/interactive/briefing/2023/10/17/israel-gaza-map-hamas-war,,,,
Unprecedented Access: Revealing and Scanning Corporate Card Expenditures of Jair Bolsonaro and Other Former Brazilian Presidents,Single project,Fiquem Sabendo,Small,"Luiz Fernando Toledo
Taís Seibt
Bruno Morassutti
Alexandre Facciolla","Luiz Fernando Toledo - Brazilian data and investigative journalist, founder of Datafixers.org. 
Taís Seibt - Director of operations at Fiquem Sabendo, journalist and PhD in Communication. 
Bruno Morassutti - co-founder of Fiquem Sabendo, lawyer, specializing in Civil Procedural Law and Public Law, and holds a Master's degree in Law. 
Alexandre Facciolla - data driven journalist, produced content and reports for various editorials.",Brazil,"In January 2023, Fiquem Sabendo achieved a historical, unprecedented and exclusive access to the physical archives of invoices detailing expenses made by Jair Bolsonaro and other former Brazilian presidents using corporate cards - which have always been sicret. The documents were scanned and made available in a digital repository by Fiquem Sabendo for public consultation. Additionally, marking a historic milestone, the Brazilian government, in response to Fiquem Sabendo's FOIA requests published, for the first time ever, a comprehensive list of expenses in an open data format for public consultation.","The data release was published in our newsletter on January 12, 2023, and immediately made headlines in the country's major media outlets. Our social media platforms experienced unprecedented engagement, and we tripled our followers within a few days. The Minister of Communications held a press conference on the same day to explain the disclosure of expenditures. The Public Prosecutor's Office requested the Federal Audit Court to investigate inappropriate expenses by Bolsonaro on the corporate card. The Ministry of Management and Innovation in Public Services opened a public consultation on a new decree to regulate card usage. Our collection of documents on Google Pinpoint was the most accessed globally. We secured support from Muckrock to create a repository on DocumentCloud. Three thousand reports citing the data were published across Brazil in six months across media outlets of all types, sizes, and regions. We reached 5k new subscribers in our newsletter days after revealing the data. Also, we won two important data journalism awards in Brazil, from School of Data (Open Knowledge Brazil) and Livre.jor. Because of our complaint the Office of the Comptroller-General ruled that these documents should be made automatically public at the end of every Presidential mandate.","Our primary resource for obtaining the data was the FOIA. The Brazilian legislation stipulates that expenditures of sitting presidents are confidential but can be made public immediately after the end of their term if someone requires it. We obtained the breakdown of corporate card expenses for former presidents, including Bolsonaro, just days after he left office. The government responded directly to our demands and made the data available proactively in an unprecedented manner. However, we went further and requested the invoices. Access was granted for the first time in history, but it required consulting thousands of physical files in Brasília, the capital. We hired freelancers and partnered with media outlets to open processes and photograph receipts, but it seemed like an impossible mission facing the volume of documents - uncountable boxes and folders. We decided to buy a scanner, hire more people, partner with more journalists, and managed to digitize about 13 thousand pages in a few weeks until the government blocked our team's access. The work needed to be supervised by a public servant to mark personal data in the processes before making them available for digitization. Then, they promised to digitize the files on their own.","The primary catalyst for our success was the ""Lei de Acesso à Informação"", the Brazilian equivalent of FOIA. The expenses of former presidents become public after the end of their term, but we realized that this transparency did not happen automatically—it required proactive requests. We strategically leveraged the law, filing our request at the opportune moment and promptly sharing the data just days after Bolsonaro left office. In terms of technology, advanced OCR and document-analysis-specific platforms were essential to make the millions of scanned pages useful. Generative AI was also essential, and we used Document Cloud´s ChatGPT Turbo feature and Microsoft Copilot to get summaries of documents that amounted to over 14 thousand pages. Finally, REGEX was necessary to find specific data that could identify ID numbers and other information required to categorize and group information.","A crucial lesson revolves around the power of collaboration in journalism. The seamless cooperation with colleagues from various media outlets played a pivotal role in unearthing compelling narratives. Without the collective effort in scrutinizing the data and uncovering stories, this groundbreaking release of information would have held little relevance in the public discourse. The project underscores the significance of journalistic collaboration as an indispensable force for driving impactful and meaningful reporting. Certainly, other journalists can use the techniques and tools for analyzing large volumes of documents that we apply in this project, as all the tools are free software or have free versions. In addition, specially Brazilian journalists and investigators, but even professionals from other countries, can use our database to discover stories, as already happened, because our collection is available for free.",N/A,N/A,Portuguese,https://fiquemsabendo.substack.com/p/exclusivo-acessamos-as-notas-fiscais,https://fiquemsabendo.substack.com/p/veja-quais-foram-os-gastos-do-cartao,https://bit.ly/3O5q8IH,https://bit.ly/3tUZVFM,,,,
Você está na rota da lama? Veja locais que seriam soterrados por rompimento de barragens,Single project,Repórter Brasil,Small,"Hélen Freitas, Marina Rossi e Hugo Nicolau Barbosa de Gusmão","Hélen Freitas é jornalista premiada pela Agência Nacional de Águas e Saneamento Básico (ANA) e finalista do Prêmio Jornalístico Vladimir Herzog de Anistia e Direitos Humanos, ambos com o especial Mapa da Água, publicado pela Repórter Brasil. Marina Rossi é jornalista premiada pela Sociedad Interamericana de Prensa (SIP) – Derechos Humanos Y Servicio a la Comunidad, com a série especial sobre o Relatório da Comissão da Verdade, publicada pelo El País Brasi. Hugo Nicolau Barbosa de Gusmão é geógrafo, possui experiência em técnicas de geoprocessamento e sensoriamento remoto, trabalhando em projetos de agricultura de precisão, mapeamentos diversos, análises de dinâmicas territoriais.",Brazil,"Oito anos após o rompimento da barragem da Samarco em Mariana (MG), que deixou 18 mortos, e quatro anos após o desastre da Vale em Brumadinho (MG), cuja avalanche matou 270 pessoas, dados obtidos com exclusividade pela Repórter Brasil mostram todas as áreas do país com risco de ficarem submersas no caso de rompimentos semelhantes.

O mapa interativo mostra que 178 municípios de 15 estados têm quase 700 áreas suscetíveis à inundação. Enfileirados, os terrenos inundáveis formariam um rio de lama com cerca de 7.000 km de comprimento, a distância de Porto Alegre ao México em linha reta.","Um ativista de Congonhas (MG), cidade destacada na matéria, contou que, após a matéria das barragens, a prefeitura instalou correndo umas placas em um dos parques do município com orientações sobre a rota de fuga. E que fariam simulados de rota de fuga no local. Diversos pesquisadores procuraram a reportagem para dizer o quão fundamental foi a divulgação dos dados para os estudos deles. O Mapa da Lama foi republicado no UOL atingindo grande audiência e reportagens de outros veículos, como o Le Monde, divulgaram o Mapa. Somente no mês de sua publicação, a Repórter Brasil teve mais de 32 mil visualizações nas páginas do mapa e reportagem.","Apesar de dois grandes desastres ligados ao rompimentos de barragens no Brasil, ainda assim as populações que moram perto de áreas de exploração da mineração não tinham acesso às áreas que ficariam submersas em caso de rompimento, ou seja, uma informação fundamental para a sobrevivência dessas pessoas ficava escondida. Além disso, a reportagem denunciou que mesmo após uma determinação da Agência Nacional de Mineração, de que as empresas devem fornecer os mapas de suas estruturas, sejam elas monitoradas ou não pela agência, com identificação das Zonas de Autosalvamento e Zonas de Segurança Secundária, nem todas enviam os dados completos. E 113 sequer apresentaram informações.","Durante 4 meses, a reportagem trabalhou em cima dos dados recebidos em formato shapefile. Primeiro, começou o desenvolvimento do Mapbox onde todas as informações seriam aplicadas e disponibilizadas para consulta pública. Após isso, fizemos o cruzamento de dados para saber o tamanho da extensão da lama e, ainda, a análise análitica da situação, selocionando quatro casos em que bairros residenciais, escolas, rios e terras indígenas correm risco de destruição nos municípios de Congonhas (MG), Brumadinho (MG), Presidente Figueiredo (AM) e Jaguarari (BA). Foram diversas conversas com pesquisadores, procurador e possíveis atingidos.","Os colegas podem aprender, primeiramente, que é preciso solicitar incansavelmente aos órgãos competentes informações de interesse público. No Brasil, temos a Lei de Acesso à Informação que nos garante esse direito e, mesmo que o órgão negue, é possível justificar o interesse público na questão, método que foi necessário para termos os dados. Também podem aprender que é possível fazer reportagens inteiras em cima de dados de georreferenciamento. Com mapas, é possível cruzar inúmeras informações como o risco para indígenas. Esse é o caso de Presidente Figueiredo (AM). Caso a barragem se rompa, uma terra indígena vizinha será atingida. Nessa história pudemos ver que a mineradora já coleciona algumas acusações de derramamento de minérios que têm atingido rios utilizados pela população indígena que ali vive.",N/A,N/A,Português,https://reporterbrasil.org.br/2023/04/voce-esta-na-rota-da-lama-veja-locais-que-seriam-soterrados-por-rompimento-de-barragens/,https://reporterbrasil.org.br/mapa-da-lama-confira-se-sua-casa-seria-soterrada-pelo-rompimento-de-uma-barragem-de-mineracao/,,,,,,
Why the haze has reached Singapore’s shores again,Single project,The Straits Times,Big,"Charlene Chua, Joseph Ricafort, Marina Rovira Boix, Roman Sverdan, Xaquin G.V
Special credits: This project was created by Joseph Ricafort, Marina Rovira Boix and Charlene Chua for their final project in the master’s programme in visual tools to empower citizens conducted by the University of Girona. Special thanks to mentors Xaquin G.V. and Matt Osborn for their support.","The Straits Times digital graphics team is a multi-disciplinary team focused on creating high-quality data-driven and visuals projects for editorial. Our audience are mostly Singaporeans from all walks of life, so we balance innovation and functionality for all of our projects.",Singapore,"Published on October 6, when Singapore's PSI reached unhealthy levels for the first time since 2019, this interactive piece explains why the haze has returned after three years and it might differ this time.","The project was well-viewed, as it was published just in time during the week where Singapore’s PSI level was in the unhealthy range since 2019. It garnered ~76k views in October alone.",We had a lot of issues with Vapor and utilised experts in the field to help create the final visualisation.,"The 3D haze model was created using a numerical simulation of the 2019 haze episode for the period from Sept 17, 2019 00:00 UTC (Universal Time Coordinated) to Sept 21 23:00 UTC. Model computations are based on the Weather Research and Forecast with Chemistry model (WRF-Chem, V. 4.2) and was kindly provided by Dr Santo V. Salinas from the National University of Singapore’s Centre for Remote Imaging, Sensing and Processing (NUS-Crisp). In this simulation, we are using hourly values of model PM2.5 as a proxy for haze and smoke particle transport. PM2.5 are particles with a diameter of 2.5 micrometres or less, and originate from forest fires or burning. We subsequently applied a colour scheme to the 3D model volume to approximate a smoke-like look.

Finally, we visualised the data using Vapor, an open source application by the National Centre for Atmospheric Research based in the United States. Special thanks to Scott Pearse, Nihanth Cherukuru and Stanislaw Jaroszynski, who kindly walked us through the process of using this platform.","Effectively communicates how smoke from peatlands in Indonesia is transported to Singapore, as what we know as Haze.",N/A,N/A,English,https://www.straitstimes.com/multimedia/graphics/2023/10/haze-returns-singapore-2023/index.html,,,,,,,