This application takes any text input and returns the most similar congressional bills using the machine learning WMD Similarity Model and Google's 2013 pretrained word embeddings
⏵ pip3 install --user --upgrade flask
⏵ git clone git@github.com:whs2k/GPO-AI.git
⏵ cd loc_challenge_aim
⏵ env FLASK_APP=challenge.py flask run
Or shortcut:
⏵ make run
- Scrape GPO data (notebooks/1.0-whs-xmlExtract.ipynb)
- Save Data as CSV (data/1.3-billTitleSponsors.csv)
- Build and Train WMD Model (notebooks/1.0-whs-xmlExtract.ipynb)
- Save Model and Similarity Instance; three files (model/-gitignored-)
- Install and Run Application