Hybrid NLP framework for enhanced sentiment analysis and topic detection on YouTube

This paper introduces an advanced hybrid NLP framework designed to enhance sentiment analysis and topic detection in YouTube ‎comments. By combining feature extraction methods like Bag of Words (BoW) and TF-IDF with neural network models like LSTM and Bi-LSTM, the framework effectively uncovers latent topics and sentiment orientations. The study specifically analyzes comments on Oscar-nominated movie trailers, demonstrating the framework's ability to capture both explicit and implicit patterns of sentiment. This study shows ‎that the Bi-LSTM model with BoW features achieves the highest performance, with accuracy, precision, recall, and F1-scores nearing 90%. ‎This hybrid approach delivers practical and theoretical developments in natural language processing applications for content creators and ‎marketers to optimize engagement strategies based on user sentiment and thematic preferences‎.