We would like to see more transparency in all levels of government. In the past few days alone there has been a cry for transparency and reform of the political system. Modern technology allows for us all to bring about such changes from a grassroots level.
What it does
Parses through all active congressional bills currently on the floor of the House and the Senate, searching for spurious language which could indicate additional legislation which is unrelated to the bill.
How we built it
Riders Block is a web-based python system using the Flask framework backed by an SQLite database. Information regarding the bills was scraped from govtrack.us using the python BeautifulSoup library. Information is then parsed using the Google Natural Language API, to generate a neural network to retrieve the textual saliency of the bill. Using the textual saliency, any spurious language can be discovered. Bills with potential riders are flagged by the front end, at the appropriate section, in a clear way for a user to interpret.
Challenges we ran into
Over the last 36 hours, a variety of challenges arose. These challenges pushed us to learn a variety of new tools and skills, but we were ultimately successful in overcoming them. These challenges were (but not limited to): - Natural Language Processing - Data Scraping from HTML - Utilization of Google API
Accomplishments that we're proud of
Jesson: "Developing technical queries and parsing it's returned value into complex data structures." Kye: "Designing and instantiating a relational database holding all active bills and their saliency" Zach: "Successfully incorporating the Google Natural Language API into the system." David: "Scraping data off of congressional websites."
What's next for Riders Block
Not all bills have a government provided comprehensive summary, without a summary we are unable to generate a saliency for them. We built out the functionality to summarize bills using the python Sumy library. The problem was Sumy did not accurately summarize the document. In the the future we would like to implement a neural network using TensorFlow to more accurately summarize the bill. This will allow our system to be able to parse all bills regardless of whether they contain a goverment summary.