The phenomenon of fake news has grown ever more rampant since the recent election cycle. What if users could, at the click of a button, determine the veracity of any news source they are viewing?
What it does
NewzCheck is a chrome extension that allows users to quickly verify any news source they are currently viewing.
How we built it
NewzCheck is built on a variety of APIs and uses the latest machine learning algorithm. We used the webhose.io API to retrieve thousands of news articles from top news sources published today. For every news article webhose scrapes, we run the text through Microsoft's Cognitive Services API to find keywords. The information for all of these articles is stored on a Microsoft SQL Server hosted on azure.
When a user presses a button on the chrome extension to verify the news source, the chrome extension pings our Django backend, hosted as an Azure Webapp. The Django backend then queries webhose to get the text and statistics on the news article. These analytics are compared to those stored on our SQL server using a machine learning classifier that we trained using a gradient ascent model.
Challenges we ran into
We ran into several difficulties using a host of APIs but luckily mentors were available to help! Thinking of training and testing data and algorithmic design for verifying news sources was also quite a challenge.
Accomplishments that we're proud of
We trained a highly accurate classifier of news. We think that there are plenty of real world applications
What we learned
We had never built an Azure webapp or Django application before so we all learned a little bit more about the difficulties of hosting backends!
What's next for NewzCheck
We plan to scale up!