In a time with increasing difficulty in discerning truth online and stakes greater than ever, we decided to try to use our various tech skills to challenge the prevalence of fake news on social media.
What it does
Sway empowers users to see the factual validity and biased slant of every article they read and share online. Our program uses machine learning to grant a score to every article on how factual and how biased it is. On Facebook, users can utilize a a Chrome extension that presents scores for news articles built into the timeline view. On Twitter, users can @ invoke our TwitterBot, which will reply with a score of factuality and bias for any given tweet.
How we built it
Challenges we ran into
We had initially hoped to be able to develop a Facebook Messenger bot that would always be listening to a user's conversation and would reply anytime some fictitious was said. However, we later learned through various methods, APIs, and SDKs that Facebook does a fantastic job securing users' private conversations, making an always-listening bot impossible. This is why we pivoted to the Twitter bot, which allows us to tackle fake news on two of the largest social media platforms today. Also, time and sleep deprivation caused many struggles.
Accomplishments that we're proud of
Ozzie tried valiantly to scrape Facebook but is most proud of his beautiful Chrome extension. Nakul was most proud of his creative backend architecture. Olivia built her first Twitter bot using Python for the first time.
What we learned
We learned that Facebook has some serious restrictions, that machine learning is complicated and tedious, and that we're all great friends. ❤️
What's next for TestProject
We'd like to add more data points to increase the algorithm's credibility.