We were interested in developing a cool web app that uses machine learning and outside sources to generate non-static realtime updated pages. Given the current political circumstances and relationship between politicians and the media, we thought that this avenue would be the perfect to apply these interests. We took inspiration in the design from old news papers, while still trying to maintain a modern aesthetic.
What it does
Newsreel uses crowdsourced data to determine potential biases of news sources in real time. In addition, it uses Natural Language Processing and twitter, a constantly updating psychological database, to learn the general sentiment about a source amongst the public. Both of these pieces of information are presented to the user along with each article so that they can understand the article and its content in greater context.
How does it relate to time?
All of our metrics, twitter source sentiment, source bias, and bias votes, are updated in real time, to give the user the best overview possible of all current events and news happening at that moment
How we built it
news div in our html, which contain the article titles, descriptions, news sources, authors, sources, bias, and sentiment. We used various APIs and libraries throughout the build of our project, most notably the
tweepy python library for gathering tweets about a certain news source, the
TextBlob python library for performing the Natural Language Processing ML, and the Google News API for gathering some news sources. We used a firebase realtime database to communicate all these metrics back and forth between our webapp and the various scripts.
Challenges we ran into
We ran into several challenges during the development of this product but the thing that gave us the most trouble was using the firebase realtime database to move data through various parts of our project. This was mainly because each language needed a different framework and had different syntax used to post and read from firebase.
Accomplishments that we're proud of
We are very proud of the fact that I designed all of the logos from scratch, (We were afraid we would not have enough time) and we are really proud of how the header and footer turned out. In addition, we are really proud of the fact that we were able to implement breakthrough machine learning and natural language processing to give
What we learned
What's next for Newsreel
We plan to implement international news that uses more machine learning to translate the English or various other languages so our users can not only get news from the 5,000+ news sources we already draw from, but also from others around the world