ClickBlock

Before
After
Summary

Inspiration

Lately Facebook posts are increasingly loaded with links and we share a mutual hate for articles that misrepresent their content in the hope of attracting user clicks. This internet trend is deceptive and tedious, so we decided to do something about it.

What it does

ClickBlock is a chrome extension that automatically detects click-bait articles in your Facebook feed. Articles identified as click-bait are highlighted in red, and have their title replaced to be a more accurate description. Additionally, the extension provides a summary of the article when the user hovers over it.

How we built it

ClickBlock was built using a flask server that handles the processing and identification of articles. A chrome extension calls the server and replaces the appropriate content.

Challenges we ran into

Initially, we tried to make calls to the server through the web page itself but ran into issues with https validation. Facebook uses https protocol and was blocking our requests to the server. To work around this issue, we delegated the calls to browser instead of the webpage.

While scraping data for training our classifier, we ran into issues with the New York times using async calls for displaying the articles. In order to solve this problem we, analysed the requests being made to identify the xhr call responsible for displaying article content, which we simulated in our scraper.

We also tried to implement our own article summarizer, but rapidly ran into issues with the quality of the summaries. Natural language processing is a large and complicated problem that we did not have enough time to solve during the hackathon.

Accomplishments that we're proud of

Our team worked together cohesively despite all of our members being from different countries.

What we learned

We learned how to create a chrome extension, as well as how to use web scraping as an effective tool for obtaining useful data.

What's next for ClickBlock

UI updates and improved article classification would provide the finishing polish for this project. Additionally, ClickBlock could transition to also grading the reliability of the news, not just the accuracy of the headlines.

Built With

Submitted to

Hack @ UCI

Created by

I worked on the server which contains the classifier and communicates with the chrome extension. I also worked on API calls for the title change and ten-line summary. I was also responsible for mining the Buzzfeed and NY Times articles that were used for the classifier training.

Ahan Mukhopadhyay
Data Scientist
Worked on mining data, and training classification algorithms with logistic regression and natural language processing

Giovanni Alcantara
Kevin Hine
Martin Georgiev
Nikita Samarin
PhD Student in Computer Science at the University of California, Berkeley