Comment Shield: read your YouTube comments without the sting

This submission is part of the Hacktoberfest Weekend Challenge, aiming to build for a friend—a YouTuber named Manasa. Comment Shield is a web application designed to help content creators manage their comments better. When scrolling through hundreds of comments, creators often come across mean-spirited remarks that can affect their mood. Comment Shield addresses this issue by categorizing comments into Positive, Negative, and Neutral.

The application was built for Manasa, a YouTuber who makes videos on Black Pink edits. Manasa shared that when she encounters several comments in the same category (positive, negative, or neutral), she tends to assume that reflects the overall opinion on her video. Comment Shield offers a more nuanced way of reading comments by allowing creators to focus on positive feedback first, while gently managing negative comments and flagging harsh ones for easy identification or skipping.

Key features of Comment Shield include:

- Positive First: The app prioritizes opening with encouraging and supportive comments.

- Gentle Mode: Negative comments are automatically blurred until the creator decides to read them.

- Harsh Flagging: Strongly abusive comments are clearly marked so the creator can easily avoid them.

- Summary Bar: Displays the percentage split of positive, neutral, and negative comments, providing a quick overview of viewer sentiment.

- Privacy: The application is designed to be privacy-first, with no backend servers or accounts. All data, including YouTube comments, process in the user's browser.

The project uses a stack that includes the YouTube Data API v3, Transformers.js for local open-weight model processing, and a small word list as a fallback if the AI model fails to load. The application is hosted on GitHub Pages and is limited to working best in English and having some potential issues with sarcasm detection.

Manasa, the creator who inspired this project, shared that Comment Shield helped her gauge her audience's favoritism and categorize comments more effectively. She noted that it reduced the time she spent manually reading comments, making her job easier. Moreover, she appreciated the open-source nature of the project, which allows for modifications and improvements by anyone interested.

This open innovation approach ensures that creators can maintain control over their comments' privacy, avoid third-party fees, and have the flexibility to adapt the project to other languages or models.