News Source Identifier

Data comes from this collection: https://www.kaggle.com/snapcrack/all-the-news/home

TODO

• What data? - gatherd from Kaggle
• How is it cleaned/prepared? - names are scrubbed from doc body and then
  tokenized
• What metrics are we using?
• What is the baseline?
• How does baseline perform?

RUN

• Save article csvs from kaggle into data/

• To run the sentiment model, you need to uncomment some lines in sentiment.py where you save the data. After saving the data, you can run python3 sentiment_model.py to run the sentiment model and get the confusion matrix.

• To run the ML + Sentiment model, cd into experiments and run python3 data_reader.py ../data/articles1.csv ... passing all articles csvs are command line arguments. This will take a while but it will generate the data and the confusion matrix.