Research On Movie Recommendation Algorithm Based On The Fusion Of User Portrait And Knowledge Map

With the development of Internet and the improvement of information technology,people have realized the freedom of information that they can know what is going on in the world without going out at home.What follows is no longer the problem of information scarcity,but the problem of how to deal with more data than people can artificially handle.The same is true for the film industry,where big data has played an important role since Netflix used data from its 30 million paying subscribers to create House of Cards.The recommendation system is an effective way to solve this problem,which can help the movie makers to find the target group conveniently,and also help the users to reduce the time spent searching for the favorite movies.Recommendation algorithms based on collaborative filtering are most often used in movie recommendation.However,the explosive growth of data in this era is both an opportunity and a challenge for recommendation systems.There are two problems in the collaborative filtering recommendation algorithm,one is the problem that only considers the user rating matrix and ignores the movie content in the collaborative filtering algorithm,the other is the problem of the sparsity of the cold start and user rating matrix.In order to solve these two problems,this paper adopts a movie recommendation algorithm based on the fusion of user portrait and knowledge map.This paper mainly includes the following three aspects:(1)In order to solve the problem of the first aspect mentioned above,this paper combines the user portrait in the recommendation system,and takes the movie content into consideration by using the characteristics and properties of the user portrait to improve the recommendation performance.By analyzing the interaction record between the user and the movie,the user portrait technology is used to accurately describe the user information,and LDA and tf-idf text mining algorithms are used to establish the text theme model of user modeling.(2)In this paper,knowledge map and recommendation system are combined to solve the problem of sparse scoring matrix.The semantic information of movies can be extracted by establishing the knowledge map in the movie field,and the recommendation performance can be effectively improved by combining with the collaborative filtering algorithm.In the process of combining the knowledge map with the recommendation algorithm,the algorithm adopted is the Ripple Net algorithm improved on the traditional algorithm,which mines the potential points that users are interested in by continuously expanding the click records of users.Finally,the recommendation algorithm based on user portrait and the recommendation algorithm based on knowledge map are combined to improve the accuracy of the recommendation algorithm.(3)Empirical research.In order to make the optimization algorithm of this paper more convincing,this paper does an empirical study on the user-based collaborative filtering algorithm on the movie data set Movie Lens,and then introduces the IMDB set to study the mixed recommendation algorithm.Finally,the recommendation effect is evaluated by the recall rate,accuracy rate,and comprehensive value F1.Through empirical proof,the algorithm proposed in this paper has a very good effect on improving the performance of the recommendation system.