Research On Hybrid Recommendation Algorithm Based On User Interest Change And Clustering

With the rapid development of the Internet,many network users are difficult to quickly obtain the information they want when they face huge network information.In order to meet the needs of users,a personalized recommendation system is generated.In the recommendation system,the collaborative filtering algorithm has been widely used.However,the user interest model established by the collaborative filtering algorithm does not take into account that user interest will change with the influence of external factors,and the data sparseness and real-time problems in the algorithm will affect the recommendation quality of the recommendation system.In order to effectively solve the above problems,this paper focuses on the three aspects of user interest capture,reducing data sparsity and improving the real-time performance of the algorithm.The main contents of the research are:1)This paper proposes a collaborative filtering algorithm based on user interest change,which is used to solve the problem that collaborative filtering algorithm can not capture user interest changes in real time.The algorithm integrates the time factor of user project scoring and the project popularity in the current time window into the collaborative filtering algorithm,uses the simulated Ebbinghaus forgetting curve function to establish the time model,and collects feedback from the user in the current time window.Quantity build project popularity model.Then combine the two models and propose a new scoring weight calculation model.Finally,the scoring weight calculation model is brought into the calculation formula of the similarity between users and the pre-scoring of the project,and the scoring in the scoring matrix is weighted.Through experiments,the algorithm can effectively capture the changes of user interest and improve the accuracy of the recommendation system.2)This paper proposes a collaborative filtering algorithm based on user preference clustering,which is used to solve the data sparseness and real-time problem of collaborative filtering algorithm based on user interest change proposed in 1).The algorithm obtains a weighted user item scoring matrix based on the algorithm in 1),and establishes a preference matrix based on the item category matrix.By introducing K-means clustering algorithm and combining preference matrix to cluster users.Use the clustered results to select the nearest neighbor of the target user and recommend the Top-N project.In this paper,the MovieLens data set is used as the experimental data.By comparing with the evaluation criteria of other recommended algorithms,the algorithm can effectively improve the real-time and recommended accuracy of the recommendation.Figure[29]table[18]reference[54].