Trusting news sources and the information we get from them has become a really big problem in the world today. We set out wanting to help improve people's understanding of the news articles they are reading and to help them identify possible sources of bias within those articles. People deserve to stay informed with accurate information and our application helps them do that.
What it does
Our chrome extension takes popular news websites, parses data from them and then uses machine learning to find out more about how the article is written. The application provides insights such as the relevant sentiment within the articles, summaries of those articles, and related articles that have different viewpoints and information, but share topics.
How we built it
Aylien API We used Aylien in order to extract the main body of news articles.
Google Natural Language API We used the GNL API to examine the sentiment and keywords within an article.
News API We used this API (which includes Google News) to find related articles via keywords.
JQuery Basic script that simplifies interaction with the DOM.
HTML/CSS Formatting and styling of the various elements within the page.
Chrome Extension API Used to do things such as save settings, interact and modify webpages, and load information from files.
Challenges we ran into
We had to make a lot of API calls so we struggled to work with the amount of calls we were provided with for each of the APIs. None of us had made a Chrome Extension before so we needed to learn how to make one together. Also the different formatting on different sites proved particularly difficult when encountering things such as fixed elements. We had to make our extension compatible with a lot of different sources.
Accomplishments that we're proud of
We learned a bunch of new technologies and were able to apply all of them to our project successfully. We ended with a clean and professional product that you wouldn't expect had started at a hackathon. We also added accessibility features that make it easier to customize the look and feel of the application.
What we learned
We learned how to use a bunch of new APIs and how to create Chrome Extensions. This project also really helped us to learn how to work together as a group over Github without interfering with one another's progress.
What's next for Purple Journal
Eventually, we will want to add more accessibility features and have certain features load faster. We tried using a fact checking algorithm, but it wasn't very accurate and didn't help that much. It would be nice to implement something like this on our own in the future. Eventually, it might be worth considering making a database to store previously interpreted data for certain sites, so that we don't have to send them through APIs every time we try access them. This will help improve loading times and API call efficiency.