Connect the Dots

Sample grouping of news articles
Hovering over an item to open the article

What does it do and why do we need it

Fake news dominate our lives today. But even if news outlets try to stay as neutral as possible, readers can never be sure that they are in fact neutral.

To alleviate this issue, we give you context for your news stories. Compare the headlines and articles of multiple sources, complete with a neutral view from Reuters. Use this to quickly inform yourself and spot fake news or sensationalist headlines!

How we built it

We collect data from multiple news outlets including Reuters, The Guardian and many others. We scrape their web pages or use the API to access the articles they post.

To match articles that belong together, we make use of the Calais API by Reuters. It allows us to find common topics between the articles and ranks them by relevance. To enhance our similarity ranking, we utilise the sklearn machine learning library. Those scores are then combined in our similarity score.

We group articles together and collect headline, image and a short summary to display on our front page. The reader simply has to open it and start reading!

Challenges we ran into

Even with the great tagging and ranking by the Calais API, some articles still don't belong together even if they deal with a similar issue. Finding out which information to prioritize was tricky.
Accessing the data of so many different sources meant that we had to find a unique solution for each one of them. A uniform API would have made this easier.