# Application name
name: Sports News

# API endpoint
api: "http://localhost:8000"

# Application layout configuration
layout:
  description: >
    # Sports News

    This application categorizes sports articles by sport, fact score and sentiment. Data is pulled via a series of
    RSS feeds. 
      
    Article titles are labeled using two categories:
      
      - Fact - hard data vs rumors
      - Sentiment - happy vs unhappy
       
    ### Tech stack
      
    A zero-shot classifier, backed by a large general language model with no labeled data, is used to categorize topics,
    score and sentiment. Additionally, a txtai index enables ad hoc similarity searches against the data.
      
    The following libraries are used:

      - txtai
      - Transformers
      - Sentence Transformers
      - Streamlit
      - FastAPI
       
    [Full source code and configuration](https://github.com/neuml/tldrstory)

  queries:
    name: Topics
    values: [Latest, --Search--, Baseball, Basketball, Football, Hockey]

  filters: [Fact, Sentiment]

  chart:
    name: Sentiment
    x: Sentiment
    y: Fact
    scale: [0, 4.5, 5.0, 5.5, 10]
    colors: ["#D32F2F", "#FF9800", "#FFEB3B", "#66BB6A", "#388E3C"]

  table:
    Fact:
    Sentiment:
      - [0, 2.0, 😠, ""]
      - [2.0, 4.0, 🙁,""]
      - [4.0, 6.0, 😐, ""]
      - [6.0, 8.0, 😀, ""]
      - [8.0, 10.0, 🤩, ""]