Research And Implementation Of Personalized Movies Recommendation System

The 21 st century is the world of network and information data flow.Since the birth of the Internet in 1969,the network has penetrated into people's lives as well as fire and electricity,and has become an indispensable part of people's lives.At present,with the development of the network,information production has become an industrial chain,and the information produced has been accumulating and expanding production,and then there is the problem of "information overload".The problem of "information overload" has affected people's daily life.As a main way of people's daily entertainment,online movie is also inevitably impacted by "information overload".However,due to the lack of accurate positioning of people's viewing needs,the contradiction between huge amount of movies resources and users' personalized needs has become increasingly prominent.Therefore,in order to ensure that people can find movies with similar interests as soon as possible,as well as movies produced by movies producers can be put in accurately,and realize the value of movies production,personalized movies recommendation system came into being.Although as early as the end of last century,personalized movies recommendation system have appeared.However,there are still many problems that cannot be solved properly.For example,the personalized movie recommendation algorithm faces the problems of sparse data,poor scalability,cold start of new users,etc.;There are some problems in the personalized movie recommendation system,such as popular pages,complex operation,and imperfect functions.The emergence of the above problems makes the current personalized movies recommendation system far from meeting everyone's needs.In view of the difficulties faced by the current personalized movies recommendation system,this thesis carried out in-depth research.First,the calculation method of user similarity is studied,a new weight coefficient is proposed,and then an improved Pearson correlation coefficient is proposed.Based on the improved Pearson correlation coefficient,an improved user based collaborative filtering recommendation algorithm,MUCF,is proposed;Second,based on the collaborative filtering algorithm MUCF,a personalized movie recommendation system is designed and developed.The system can provide users with: movies appreciation,retrieval,movies scoring,movies recommendation and movies evaluation services.The main work of this thesis is as follows:(1)An improved user based collaborative filtering recommendation algorithm,MUCF,is proposed.The traditional Pearson correlation coefficient only considers the items that users evaluate together.Once the data is too sparse,it will affect the calculation accuracy of user similarity.In view of the above problems,this thesis improves the traditional Pearson correlation coefficient.A new weight coefficient weight is proposed,and an improved similarity calculation method is obtained by combining the advantages of Pearson correlation coefficient,jacquard coefficient and weight coefficient.Based on this improved Pearson correlation coefficient,we further propose an improved user based collaborative filtering recommendation algorithm,MUCF.Experiments on the Movie Lens dataset show that the MUCF algorithm can achieve better recommendation performance.(2)According to the principle of software engineering,this thesis analyzes the demand of personalized movie recommendation system.First,the design goal of the whole system is proposed;Second,the feasibility of the system is analyzed to ensure the normal operation of the system;Third,the top-level use case analysis of the system is carried out to clarify the role identity;fourth,the functional modules of the system are divided to meet the needs of users;Finally,the performance requirements of the system are analyzed to ensure that users can have a better user experience when using the system.(3)On the basis of demand analysis,this thesis designs the overall architecture,functional modules and database of personalized movie recommendation system.The overall architecture of the system adopts B/S structure and Django as content management framework,with clear functions of each layer.Through the functional module design,the whole system is divided into several low coupling,high cohesion subsystems.We analyze the design method of each module in detail.In addition,the system automatically generates all the table structures in the database through Django.(4)Based on the analysis and design of the system,the personalized movie recommendation system is further implemented.Based on the development language python,web application framework Django and MySQL database,the development of personalized movie recommendation system is completed.The system adopts the improved collaborative filtering algorithm MUCF proposed in(1)to provide personalized movie recommendation for users.In addition,we have carried on the detailed function test and performance test to the system respectively,thus ensuring the correctness of the system.