We hate pedophiles. We decided to find a way to spot them online using the help of some fancy algorithms, AKA. AI. In an Orwellian fashion, it boasts a Social Score (or Suspicion Score) rating, in other words 'judging' people. This project also shows how much data can potentially be retrieved with basic permissions from a Facebook App.
What it does
It scrapes the data of suspicious user-supplied facebook profiles using the Facebook Graph API, parses the data, and runs through some algorithms to determine whether the profile is legitimate, or potentially a predator (or Chris Hansen) in disguise.
How we built it
With Golang, a web server, hatred, tiredness, caffeine, and a bit of fairy dust.
Challenges we ran into
Literally anything to do with APIs and parsing data (damn strong-typing...) Making a site with Go that looked pretty was also pretty hard and figured I couldn't learn it literally overnight.
Accomplishments that we're proud of
A website that has a demo of what we tried to accomplish (as well as the backend source we would have used in the case of everything working.
What we learned
Don't trust social media. Don't trust anyone online. I'll stick to using a simple TCP/UDP stack in Go rather than building an HTTP server.
What's next for WebWatchers
Fixing the backend code, finding a way to dynamically interface with the frontend to not only retrieve OAuth tokens from a user, but get more permissions from Facebook Graph API and find data related to other people (without needing their OAuth.) It's a long uphill battle.