Design And Development Of Recommendation System Based On Message Passing Algorithm

With the development of the internet technology,the amount of information has exploded especially the development of mobile networks.It is very urgent to provide accurate and personalized services for users in a huge amount of information.A proposed algorithm is developed to solve this problem,the recommended is raised.With the special needs of users and the emergence of new projects,the traditional recommendation algorithm has many new problems,such as cold start,insufficient recommendation,and sparse data.Faced with these problems,many new technologies have been applied to the recommendation system,resulting in a number of new recommendation methods,such as deep learning-based recommendation methods,collaborative filtering-based advancement methods,matrix-based decomposition-based recommendation methods,and so on.However,these methods have lower accuracy in the absence of user information.This dissertation proposes a recommendation method based on the message passing algorithm.The message passing algorithm can compute the belief of each node.The algorithm is an information iterative algorithm based on factor graph model.And the algorithm has been applied to various fields,such as combination optimization,multi-objective linear programming,image combination and so on.The user's edge probability value can be obtained to recommend the user.Specifically,this dissertation first crawls the information of the Douban movie network,preprocesses the crawled data,and performs Chinese sentiment analysis on the movie review data.Finally,the integrated score of the film is obtained by using the built movie score synthesis model.Based on the comprehensive scoring,a factor graph model is constructed,and then the message passing algorithm is solved on the factor graph model.The effectiveness of the algorithm is verified by comparison with other recommendation algorithms.The recommendation system was built by Python technology,and the recommendation algorithm designed in this dissertation was embedded into the system.Finally,a film recommendation system based on flask was implemented.