Research And Realization Of The Personalized Movie Recommendation Engine

With the continuous progress and development of Internet technology,it's easy for user to access information.It also brings the problem of information overload.Compared with the search engine,personalized recommendation technology can help users to find their real interest information in the massive data to alleviate the information overload problem more effectively.In order to let users get rid of the limitation of massive movie data,a collaborative filtering recommendation improvement algorithm based on probability matrix factorization model was proposed.Besides,we designed and realized a personalized film recommendation engine which is based on B/S mode MVC three-layer framework by this improved algorithm.Our main work has done as follows:First,the recommendation algorithm is the most important part of the personalized recommendation engine.This paper introduced and discussed three kinds of recommendation algorithms,which are content-based filtering,collaborative filtering and hybrid filtering recommendation algorithm.And then we compared the advantages and disadvantages of these different classes of algorithms.These algorithms have the problem of sparseness,cold start,poor scalability and poor diversity and so on in application.Besides popular ranking recommended algorithm alleviated the cold start problem.Second,we used the statistical method to extract contextual features with significant difference,which reduce the dimension and sparsity of the data.Then we incorporated context correlation by calculating adjusted cosine similarity into probability matrix factorization recommendation,then an improved algorithm was obtained.The improved algorithm was tested on three different types of data sets.The experimental results showed that the recommended accuracy can be effectively improved by the algorithm.Finally,we analyzed the functional and non-functional requirements ofthe system.And we designed the overall architecture of the personalized movie recommendation engine,which is mainly based on the B/S mode MVC three-layers framework.This framework helped to develop system quickly.The we designed the database table and system components which mainly included off-line computing module,online recommendation module,movie information management module and user information module.By using the improved recommendation algorithm mentioned above and the Django framework in Python Web technology,the above functional modules can be realized.So probability matrix factorization recommendation model can be regularly trained offline and online movie recommendation service can be provided in real time.