YouTube Sentiment Analysis
Chrome Extension
Overview
A Chrome extension that performed real-time sentiment analysis on YouTube video comments. The extension analysed comments and classified them as positive, neutral, or negative, providing visual statistics and an interactive model training interface.
How it Worked
Comment Scraping — Fetched YouTube comments via the YouTube Data API v3.
Tokenisation & Processing — Processed comment text through a custom NLP pipeline.
Sentiment Classification — Applied a trained sentiment model using words, bigrams, and trigrams with associated scores.
Visual Display — Injected sentiment results directly into the YouTube page with overall scores, positivity percentages, and filtering controls.
Interactive Training — Users could manually adjust sentiment classifications to improve model accuracy over time.
Technologies
YouTube Data API v3
Fetched video comment threads in batches for processing.
Chrome Extensions API
Background and content scripts for seamless browser integration.
IndexedDB
Persisted the sentiment model locally for offline use.
Custom NLP Pipeline
Word, bigram, and trigram-based sentiment scoring with user feedback loop.