Web-based Sentiment Classifier for YouTube Comments

The current paper presents a web-based program that will evaluate the comments posted on YouTube on the basis of an analysis procedure that utilizes sentiment classification methods supported by contextual information gained through video transcripts.The system will first of all extract the comment, then, semantic search is performed to retrieve the most relevant transcript parts, and then the particular aspect is identified to which the user is agreeing, disagreeing, or assuming a neutral situation.A retrieval-augmented plan is implemented together with a language model, which adds the sentiment predictions with a human-readable explanation.This will increase the credibility of interpreting sophisticated remarks, especially those that may mention a particular spoken material in the video.The system aims at backing content creators, moderators, and viewers by providing a more subtle and dependable sentiment analysis.