Research On Movie Recommendation System Based On Sentiment Analysis

Nowadays,it is an information society.Under the background of a large number of data,data mining has become the mainstream research direction of large companies.In various industries,large-scale data often makes people unable to make effective choices.When selecting movies,users can’t quickly choose their favorite movies,or some users don’t know what kind of movies they want to see.Therefore,recommendation system has become an indispensable existence in movies.The system only considers the key factors expressed by users in the classification of comments.At the same time,it can only give a score to other items through the classification of users’ preferences in the traditional recommendation algorithm.The use of emotional dictionary can effectively analyze the text.However,this method depends on the perfection of the dictionary.The establishment of emotional dictionary in professional field can greatly improve the accuracy of text analysis.In view of the above problems,the main work of this paper is as follows:(1)Firstly,the existing methods of constructing emotional dictionary in professional field are studied,and the emotional seed words are used to fuse word2 vec and point mutual information.In terms of emotional seed word extraction,this paper analyzes the problem that the traditional method does not consider the distribution of feature words in different types of text and word position in text level text,and puts forward an improved TF-IDF algorithm,which improves the accuracy of keyword extraction.At the same time,it also proves that the performance of this dictionary is better than other dictionaries in film.(2)For the problems existing in the traditional recommendation algorithm,the possible improvements in the similarity formula are studied.The traditional method only considers the scoring relationship of user items,and does not consider the number and time factors of user scoring,so this paper will score The quantity weight and time weight are introduced into the Pearson formula,which is combined with sentiment analysis,and an improved similarity calculation method based on sentiment analysis is proposed.Analysis of traditional algorithms.(3)Considering that the method proposed in this paper needs to have practical significance,the Springboot framework is used to design and develop a movie recommendation system.The design process includes database design and system module design,and the improved algorithm in this paper is embedded in the movie system.Through the visual interface Show it to the user.