Comments Polarity Classification Based On Emotion Analysis And Movie Recommender System Design And Implementation

With the development of information technology,Internet application gradually infiltrated in all aspects of people’s life. Film, as a kind of important modern leisure entertainment, its route of transmission is also presented the characteristics of diversification. The current mobile network rapid development, make people have a wide variety of choice, watching movies online also gradually upward trend.How to identify the excellent movie on the network and recommend has become a hot research, movie recommendation system has become an important helper to solve this problem.Traditional method is the use of movie recommendation target customer and his scoring record to find similar users, using the user’s interest in films or find top-n and it is similar to a movie do recommend, however, for users to have a film project ratings is based on their scoring criteria, everyone rating scale is inconsistent, leading to unsatisfactory results. However, people often ignore the user viewing the contents of the movie subjective evaluation. So, how sentiment analysis is introduced into the movie recommendation system will provide new ideas for the development of the recommendation system. Existing sentiment analysis is broadly divided into two categories, dictionaries and machine learning method based on the existing sentiment analysis algorithms have advantages and disadvantages, resulting in a single sentiment analysis method can not be applied to movie recommendation system, reducing the system performance. So, how will the combination of sentiment analysis and recommendation to provide users with an intuitive, accurate feedback of movie viewing, allowing users to adopt the correct recommendation result in system implementation has become an important issue.Through the proposed fusing emotion word frequency and theme dimension extension can classify on comments in the system,the user can intuitive understanding of the others feedback.Add it to the system and join the "rate" formula achieved popular recommendation module.Specific work is as follows:1. Based on the ASP.NET platform using C# language to implement a recommendation movie recommendation system, user functions include Hot, personalized recommendation, high score recommend, comment polarity classification, making friends in the web.2. This paper presents an emotional frequency method fuse topic dimension extension, to use the topic dimension and emotional frequency to extraction feature and LIBSVM to classification.We can automatically determine the polarity of emotional comments,allowing users to find other users more intuitive for a movie feedback. By fusing the emotional words frequency and the topic dimension extension method, this dissertation introduces the emotion information of the comments to the recommendation model, and then applies the modified model to movie recommendation system. Experiments on Douban dataset have illustrated that, our method has better performance on F1, precision and Recall.3.The popular recommendation module has implementation by the proposed algorithm combined with content-based recommendation, high score recommended.The three synergy, effectively solve the problem of the system of cold start, makes the system has general applicability to the users.The system Add management module, it let administrators manage movie data effective, convenient to update timely.