HN - frenly | Devpost

frenly™ - Building a friendlier web community

Keywords: Internet Toxicity; Sentiment Analysis; Chrome Extension;

Inspiration

The web is just toxic because you and I can be toxic.

The Problem of Internet

Everyone hates that the internet can be really mean.
Everyone hates the cyberbullies, the keyboard warriors, the irrational haters that contribute to this issue.
Yet, everyone can easily become a bully, a hater, a keyboard warrior because people may get angry and act on anger without thinking twice.
Hence, everyone is in this vicious cycle of internet toxicity. Creating it, spreading it and hating it all at the same time.

As a conclusion, the internet needs to have a sort of measurement, that could not only think twice for us but also help us to think twice.

We need a “frenly™” internet.

Solution

frenly™ is a Chrome extension that vows to end the vicious cycle of internet toxicity in the web, by proactively preventing toxicity from spreading and creating. Features as below:

1. Read Post / Comments

When our backend sentiment analysis engine detects potentially toxic comments in the loaded web page, they will be censored by an overlaying color block.
The user may choose to reveal the negative comment at his/her risk.
If the page contains too many negative comments, we will request the user to consider to leave the page.
By smartly detecting the web content, user can be in a more active position countering internet toxicity. Whatever is spread, can no longer harm.

2. Write Post / Comments

We introduce a toxicity meter on the comment textbox. Before a comment is being posted, it is sent to our backend sentiment analysis engine to evaluate its polarity.
If the polarity goes too negative, the system will pop a window to advise the user to reconsider his/her word choices.
By initiating respectful reminding alerts, user can no longer fall easily for the emotional trap. Whatever is toxic, can no longer be created.

3. Post / Comment Wrap-Up

A health check summary is provided to the user to assess his/her average negativity and subjectivity of input history.
Implementation wise, we pull the active user's comments and posts using Facebook API and Twitter API and feed them into our sentiment analysis engine for evaluation.
Tokenization is performed on the user's comments and posts to find out the top 5 most significant words used in each social media platform.
By actively studying user’s internet speech, user can learn more about him/herself and cultivate the right mindset.

Together with all 3 aspects, we hope to create a lightweight, yet powerful tool to help to build a nicer, friendlier web community.

What's Next For frenly™

frenly™ is now supported to run on Facebook, Twitter and Youtube. We hope that it could be extended on other social media platforms to continue its effort on combating cyber negativity.
frenly™'s backend sentiment analysis engine could be improved to achieve more precise detection on negative and potentially harmful comments.