As high school seniors, we are all taking a government course this year, and a key part of the class is staying up to date on current events in politics. However, in recent years, with the country being more politically polarized than ever, it's not always easy to find an unbiased source of news. On top of this, fake news has become increasingly prevalent, especially in a world where much of the media we consume is online. However, one day during a discussion of current events in class, our vice principal interrupted to share that he used several apps to remain well educated about American politics and policy. He had anything from Breitbart to MSNBC, and numerous other news applications for the sole purpose of receiving a mixed pool of ideas and viewpoints, and after hearing from all the sides he was able to develop his own informed beliefs. This practice became the inspiration for Debrief.
What it does
Debrief is a news aggregator and analyzer. The application is entirely independent, capable of tracking current trends, and retrieving and analyzing news articles related to these topics. At its core, Debrief uses cutting edge natural language processing technology and machine learning to discover the sentiment behind each article and locate a wide set of key words and entities. Using these, Debrief builds summaries of each article and collects them all under one roof to be presented to to the reader (thats you!). On our website where the news is listed, we also have visual aides (like graphs) to display the variety of news each source produces, display trends between multiple providers, and help you to see which articles will provide the most diverse viewpoints.
How we built it
We created a Node.js server that pulls articles about politics from a variety of sources, finds overarching trends, and performs natural language processing with the help of Google's Cloud API. These articles are assessed for sentiment toward people, places, and things in the article and are shown in the website in a way that readers can easily which news sources are biased towards which things.
Challenges we ran into
Challenges we ran into was how to show to a common reader the difference in sentiment expressed by the different articles. We overcame this challenge by creating helpful graphics that explained the difference.
Accomplishments that we are proud of
We are very proud of creating an application that we think is very topical in today's political climate. With so many sources of news and many different perspectives, we think that this website will help some Americans become more informed about politics.
What we learned
We learned a lot about natural language processing and how to make interesting graphics that would appeal to and inform the common reader.
What's next for debrief
Next for debrief is to take our natural language processing even further and create totally unbiased news by synthesizing the articles from our many different sources.