Design And Implementation Of Movie Recommendation System Based On Hybrid Recommendation Algorithm

With the progress of social technology and the rapid development of the internet industry,the era of big data for information overload has arrived.As one of the methods to solve information overload,recommendation technology has been integrated into people’s lives.Since its inception,recommendation algorithms have not been flawless.Traditional recommendation algorithms have some shortcomings,such as data sparsity,cold start,and low recommendation accuracy.In order to reduce the shortcomings of these problems on recommendation systems,researchers and experts have conducted different studies to improve and optimize different algorithms,making up for the shortcomings of the original algorithm.In this paper,data sparsity and low recommendation accuracy in traditional collaborative filtering algorithms are studied accordingly.The main research contents and contributions of this paper are as follows.(1)Through research,it was found that popular movies are often recommended to users frequently in some movie recommendation systems,which affects some users’ selection of non popular movies.Therefore,this article adds a penalty term to the calculation of user similarity to solve this problem;The traditional collaborative filtering algorithm does not consider that the user’s interests will change with time,so this paper adds a time factor in the calculation of user similarity to alleviate this phenomenon.The experimental data shows that compared with the traditional collaborative filtering algorithm,the Precision and Recall values of the improved algorithm in this paper have increased by 6% and 5%,and the MAE value has decreased by 3%～4%.(2)Research has found that traditional matrix decomposition algorithms often have poor recommendation performance due to sparse matrix data.Therefore,this article improves the matrix filling method in matrix decomposition algorithms.Firstly,the optimized SVD algorithm is used to fill the original user rating matrix for the first time,and then a new matrix is constructed by incorporating project attributes.Finally,the optimized slope one algorithm is used for secondary filling.Compared with the traditional collaborative filtering algorithm,the precision and recall values of the algorithm after the matrix filling method is improved by about 10% and 11%～12%,and the MAE value is reduced by about 4%～7%.(3)In response to issues such as poor clustering performance of a single Kmeans clustering algorithm,this article combines the K-means clustering algorithm with the mouse swarm optimization algorithm to obtain a fusion of K-means clustering algorithm and mouse swarm optimization algorithm(KRSO).Compared with a single K-means clustering algorithm,the fusion of KRSO algorithm in this article has better clustering performance.This paper combines the advantages of KRSO algorithm and Light GBM algorithm in machine learning,mixes the integrated machine learning,collaborative filtering algorithm and matrix decomposition,and constructs a hybrid recommendation algorithm.Experiments show that compared with a single algorithm for recommendation,the hybrid model recommendation effect is better.Compared with the traditional collaborative filtering algorithm,the precision value of the improved algorithm in this paper is increased by about 15%,and the recall value is increased by about 18%,The MAE value has decreased by about 10%.(4)This article designs and develops a movie recommendation system based on Django,which implements the relevant functions of administrators and users.During the development process,it was found that some scholars overlooked the security issue of user passwords.Therefore,this article added RSA algorithm to protect user passwords when logging in,further ensuring the information security of users.