Hybrid CNN LSTM Approach for Sentiment Analysis of Bengali Language Comment on Facebook

This study addresses the challenge of sentiment analysis in Bengali Facebook comments, where users express their opinions on various topics. Existing sentiment analysis techniques have difficulty achieving high accuracy in this situation, which has inspired the creation of a fresh methodology. To improve sentiment analysis of Bengali Facebook comments, we suggest a model that combines a hybrid convolutional neural network (CNN) and a long short-term memory network (LSTM). Our research adds to the expanding area of sentiment analysis, especially for languages with few resources, such as Bengali. The success of the Hybrid CNN LSTM model opens the door for more precise and context-aware sentiment analysis on social media platforms, allowing for a better understanding of user sentiments and boosting user experiences. Our study focused on dividing comments into five groups according to their emotional content: troll, not bully, sexual, religious, and threat. We carefully generated the handmade dataset and ensured the quality and relevance of a dataset of 50,000 Bengali comments before using it to train and test our algorithm. The Hybrid CNN LSTM model outperforms conventional approaches and achieves an outstanding accuracy rate of 92.09%, demonstrating amazing efficiency in sentiment recognition. This dramatic increase in precision indicates that machine learning techniques are successful in sentiment analysis of Bengali Facebook comments.