# CPV Classifier

[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.6554604.svg)](https://doi.org/10.5281/zenodo.6554604) [![Project Status: Active – The project has reached a stable, usable state and is being actively developed.](https://www.repostatus.org/badges/latest/active.svg)](https://www.repostatus.org/#active)

In this work we compare different approaches to Common Procurement Vocabulary (CPV) codes classification, using data extracted from the [Spanish Treasury](https://www.hacienda.gob.es/es-ES/GobiernoAbierto/Datos%20Abiertos/Paginas/LicitacionesContratante.aspx). 

## Resources
All training, testing datasets and code are [available in Zenodo](https://doi.org/10.5281/zenodo.6554604). Check them out!

## Pre-processing
In order to preprocess the data, follow the steps below:

1. Download the atom datasets into a folder (in our case `licitacionesPerfilesContratanteCompleto3_2019`)
2. Run `atom2csvBS.py`, which will save the results in `docs.csv` 
3. Run `data2tt.py` using `docs.csv`. As a result, you will obtain `train.csv` and `test.csv`, which are the datasets used in the notebooks

## Notebooks
The notebooks below contain the different approaches compared:

Test for classic approaches: https://www.kaggle.com/marianavasloro/cpvtwodigitsml

Test for the MKaan approach:
https://www.kaggle.com/code/marianavasloro/mkaan

Training and evaluation of fine-tuned RoBERTa:
https://www.kaggle.com/code/marianavasloro/fine-tuned-roberta-for-spanish-cpv-codes


## Datasets

Full: https://www.kaggle.com/datasets/marianavasloro/dataset

10%: https://www.kaggle.com/datasets/marianavasloro/dataset10


## How to cite
If you use these datasets or our notebooks, please cite this repository and our 2022 SEPLN paper (to appear):

```
@article{PLN6429,
	author = {Maria Navas-Loro y Daniel Garijo y Oscar Corcho},
	title = {Multi-label Text Classification for Public Procurement in Spanish},
	journal = {Procesamiento del Lenguaje Natural},
	volume = {69},
	number = {0},
	year = {2022},
	keywords = {},
	abstract = {Public procurement accounts for a 14% of the annual budget of the different governments of the European Union. In Europe, contracting processes are classified using Common Procurement Vocabulary codes (CPVs), a taxonomy designed to facilitate statistical reporting, search and the creation of alerts that can be used by potential bidders. CPVs are commonly assigned manually by public employees in charge of contracting processes. However, CPV classification is not a trivial task, as there are more than 9,000 different CPV categories, which are often assigned following heterogeneous criteria. In this paper we have created a CPV classifier that uses as an input the textual description of the contracting process, and assigns CPVs from the 45 top-level CPV categories. We work only with texts in Spanish, although our approach may be easily extended to other languages. Our results improve the state of the art (10% F1-score improvement) and are available online.},
	issn = {1989-7553},
	url = {http://journal.sepln.org/sepln/ojs/ojs/index.php/pln/article/view/6429},
	pages = {73--82}
}
```

## Acknowledgements
This work has been supported by NextProcurement European Action and the Madrid Government (Comunidad de Madrid-Spain) under the Multiannual Agreement with Universidad Politécnica de Madrid in the line Support for R&D projects for Beatriz Galindo researchers, in the context of the V PRICIT (Regional Programme of Research and Technological Innovation). 

We would like to thank **Jennifer Tabita** for her contributions to the initial set of notebooks, and the AI4Gov master students for their validation of the approach. 

Source of the data: [Ministerio de Hacienda](https://www.hacienda.gob.es/es-ES/GobiernoAbierto/Datos%20Abiertos/Paginas/LicitacionesContratante.aspx).