TRST Real Story Tester

Chrome extension on page with low trust rating
Chrome extension on page with higher trust rating

Inspiration

The modern world is inundated with false news stories, breaking up democracies and fracturing society.

What it does

It uses natural language processing to build similarity indices between news articles found under the keywords built from the article content. These are collated from trusted sources, and a heuristic forms an overall index for the articles trustworthiness based on its similarity to articles from trusted sources, and the number of sources found, if any, that also discuss the article. Factors such as the recentness of the article and other associated information are also taken into account. The system is designed to be easily accessible through a Chrome browser plugin and a interactive website, both of which provide an easy to use, simple, clean interface to display relevant information. We use a simple colour score for the extension icon to passively inform users of the trustworthiness as they're reading their articles, allowing it to not interfere with the browsing experience, whilst allowing them to click for an expanded information panel for more detailed information.

How we built it

We process the articles through...

The web scraping for article collection was built with ...

The web server is built on top of Flask, a web framework for Python. In

The Chrome extension is built with JavaScript and JQuery with a HTML/CSS frontend. It uses the Chrome extensions system to integrate directly into the browser for an interactive end-user experience. The website is built...

Challenges we ran into

Natural language processing finding links between articles proved difficult as... . It was also a challenge to come up with a mathematical model of reducing the similarities to single correlation factors.