In today's polarized political climate, many don't venture out of their social groups, which makes even hearing opposing viewpoints difficult. My team at HackDavis 2017 and I wanted to create something that would help people find both sides of an issue, because hearing those we disagree with is the first step towards getting along.
Perspecticles is a Google Chrome extension that subtly highlights people, words, and phrases that are particularly controversial. You simply move your mouse over the highlighted area, and a pop-up will display two links to one favorable and one unfavorable news article regarding that topic.
We used the IBM's Watson and Bing's search engine APIs to curate news articles, giving them a favorability rating with IBM Watson's Alchemy Language natural language processing. A sample of articles, as well as their sources, is shown above. Watson was able to detect how each article presented the topic at hand, so that our extension can select which articles should be shown under "favorable" and "unfavorable."
I was responsible for creating the backend of Perspecticles, which meant collecting articles from Bing, sending them through Watson, and caching the articles that Watson found particularly one-sided. I also created the endpoint that the extension could access to obtain an article when a controversial word appeared on a page. To do so, I used Django, as it allowed a very simple and aesthetically-pleasing way for us to visualize the data as we were working with it.