Does thinking about how much data Facebook has on you terrify you? Do you desire to make yourself more data-concious but fear what you may find? Then look no further and come play FaceJeopardy! Like normal Jeopardy, but based on your everything about you!
What it does
First, download your personal data from Facebook. Then, run it through our application, and enjoy a game of Jeopardy! that's based off everything Facebook knows about you.
How we built it
- The parser is loads each relevant JSON file and extracts the needed information.
- The natural language processor re-formats the given information into Jeopardy! style questions.
- The post processor selects the most Jeopardy! like question from each tidbit of given information/
- The application script takes the post-processed questions and generates a Jeopardy! board for gameplay.
Challenges we ran into
- Neither of us had worked with JSON files before, and parsing took longer than anticipated
- Configuring our computers to run on Python 2.7 (what was required for the NLP parts of the code) was a big pain (Isn't it always?)
- We ran out of time and ended up pulling a basic app from Github instead of trying to learn to make it ourselves.
Accomplishments we're proud of
- We were able to work out all the issues and successfully parse all the relevant JSON files.
- After finally configuring our environments, we were able to use the NLP programs relatively easily.
- Another neural network to decide which clue is the most Jeopardy! like, instead of our manual rules.
- Context is something that's missing from these questions - for example - "Facebook knows you like this" is a much harder clue to guess than "Facebook knows you like this school, known for its husky mascot". A great way to improve this project would be to do some sort of web-scraping on the extracted data to generate clues with more relevant context.
- Making this a webapp instead of a Python app.