Research on Recommendation Methods Based on Sentiment Analysis and BTM Topic Modeling

With the rapid development of the Internet and e-commerce, the importance of recommendation algorithms has become increasingly prominent. Data sparsity and scoring dependence are problems with most current recommended algorithms. In this paper, we propose the recommendation model of SABTMCF (sentiment analysis and BTM collaborative filtering). Based on the traditional collaborative filtering algorithm, the sentiment analysis and BTM topic model are used to mine the review data to obtain the user's real potential emotional emotions and different attributes of the product. The scoring matrix of the theme can alleviate the above two problems, and then calculate the similarity of the user's emotional preferences to construct the recommendation model. The paper uses Dangdang's comment data set as experimental data, and the results show that the SABTMCF algorithm can improve the data sparse problem to a certain extent and has better recommendation accuracy.