Real-Time Comment Filtering: A Chrome Extension for Toxicity Reduction on Facebook

The work aims to design a model which conceals the negative comments or bad comments from the Facebook site from the user end side. An extension is designed based on hiding the negative comments from the Facebook in which that extension using a dataset of about 2000 sample which contains positive, negative and neutral comments. The comments are fetched from the random Facebook post. The basic functionality of the extension is achieved by having both content and background scripting combined. The dataset is stored locally in the background script using Chrome's storage API once it's been loaded from a JSON file. Since the comment sections in Facebook are dynamic, a Mutation Observer monitors them to ensure newly loaded comments are also filtered automatically. The experience is enhanced and less intrusive by the content script, which compares a comment to the dataset and eliminates those rated as unfavorable by hiding or dimming neutral ones. In this regard, the extension also features a debounced filtering technique that reduces unnecessary function calls with repeated DOM changes. Such enhancements in the future, including the addition of advanced machine learning models as well as the capability for advanced sentiment analysis, are facilitated through this methodology. This project is a significant step towards creating a more moderated and controlled social media network, which will ensure a healthier online interaction, by having a scalable and adaptable framework.