Inspiration
As we all know internet a.k.a. world-wide-web is full of hatred, homophoby, racism, negativity and so on... And that's why we've come up to a idea on how to make people be aware of sensitive text all across the internet.
What it does
Our Application is in form of Google Chrome extension that runs on background and scans your whole web application for sensitive content and neatly blurrs it. No need to be afraid, you can always view what's behind the blurred curtain simply by hovering over the censored text. Than you have also available Hate Speech Dashboard. The Dashboard visualizes the censored hatred from all across your search history and create positive to negative ratio 24/7, even throughout the day.
How we built it
As it was said earlier, we used Google Chrome extension built in JavaScript for non-stop scanning pages, and scraping sentences for backend to evaluate their weight. Collected data is processed by our neatly written backend. Backend takes all the scraped text sent by extension and uses Machine Learning API from Microsoft to detect the negativity and positivity in the sentences. Based on the results, data is sorted and stored in SQLite database. Last but not least comes the dashboard. Dashboard is written in React framework as web application. Our dashboard takes stored sentences and their weight from every page and dynamically creates component individually for every visited page. The dashboard can be accessed here: https://testtestetst.tech
Challenges we ran into
One of the greatest challenges was to learn how to use Google Chrome extension. This was the first time we've had the honor to work with google extensions. It was new learning experience and that's what makes it challenging.
Microsoft ML API wasn't a piece of cake either. We had problems like slow evaluation and API crashing. Connecting the database to our extension was kind of a problem. Any of the requests from our frontend was not working. Thankfully after full focus of the whole team, the problem was solved.
Unexpectedly Dashboard functionality was not that big of a problem as the design. We wanted to create the most intuitive and clean looking design as we could while preserving functionality, and it was not a easy job to do.
What we learned and Accomplishments that we're proud of
As we said earlier, we learned to use Google Chrome extensions (that was the greatest accomplishment for all of us). Deployment of web application and databases to production. Use of custom domain etc...
What's next for HateBlurr
User authentication for personalized evaluations. More reliable Machine Learning model for more accurate hate speech evaluation. Optimalization for smoother data loading in dashboard and page scanning by our google chrome extension. Some pages wont get blurred because of the database being overloaded. Backend needs more optimization.
Log in or sign up for Devpost to join the conversation.