Sentiment Analysis With Youtube Comments Using Deep Learning

Abstract

In the modern digital ecosystem, YouTube serves as a
dominant platform for sharing content and engaging
with audiences worldwide. Every video accumulates a
vast number of comments that reflect user opinions,
sentiments, and reactions. Manually interpreting this
data is challenging due to its volume, diversity, and
unstructured nature. This project, YouTube Comment
Classification, introduces a web-based solution that
automates sentiment analysis of YouTube comments
using Natural Language Processing (NLP).
The system enables users to input a YouTube video URL
and automatically fetches comments through either the
Google YouTube Data API or a fallback mechanism
using Pytube and web scraping. These comments are
then processed using TextBlob, an NLP tool that
evaluates each comment’s polarity and subjectivity.
Based on set thresholds, comments are categorized into
three sentiment classes: positive, negative, or neutral.
The frontend is developed using React.js with Vite,
styled using Tailwind CSS, and incorporates Chart.js
for interactive data visualization. The backend is built
with Python Flask, which handles request processing,
comment extraction, and sentiment classification. The
application supports responsive design and includes
dark mode for better user experience.
This tool provides a practical and scalable approach for
understanding public sentiment on YouTube content,
offering value to creators, brands, and analysts. With
real-time analysis, fallback reliability, and visual
summaries, YouTube Comment Classification
demonstrates the power of combining modern web
technologies with NLP to deliver meaningful insights
from social media interaction