Research On Collaborative Filtering Recommendation Algorithm Based On Item Hierarchical Agglomerative Clustering

With the rapid development of information technology, and the wide spread of Internet applications, more and more information emerged on the Internet, and the growth rate of information is far beyond the capabilities that people can receive. Oriented to the problem of information overload, it is difficult for people to efficiently find the information that meet their requirements. To solve this problem, personalized recommendation technologies have emerged and continue to develop, which have greatly promoted the application of personalized recommendation system, and can provide targeted, personalized recommended service according to user preferences. At present, the collaborative filtering algorithms in personalized recommendation system are widely used, especially in the area of electronic commerce. Traditional collaborative filtering algorithm have problems like data sparsity, cold start, or low scalability, bad real-time performance, and low recommendation accuracy. If we can effectively solve the problems, the customer satisfaction degree can be greatly improved.Through the in-depth study of collaborative filtering recommendation system, collaborative filtering system has been wildly used currently. In consideration of datasparsity,badreal-time performance and low recommendation accuracy and other problems, this paper put forward collaborative filtering recommendation algorithm based on item hierarchical agglomerative clustering. First of all, in view of data sparsity, through considering item description similarity and item functional similarity complexly, we weight item description and item functional similarity to obtain item characteristic similarity, then set characteristic similarity matrix and cluster the item. In addition, in view of low recommendation accuracy and bad real-time performance, at the same time considering k-means clustering algorithm to select the initial center improper and the change of cluster centers will bring the problem of high computation time complexity, we use agglomerative hierarchical clusteringalgorithm to improve and propose agglomerative hierarchical clustering algorithm based on item. Finally, in order to improve the recommendation accuracy, in view of the problem of high similarity calculation in collaborative filtering recommendation, we introduce regulators to improve rating similarity calculation and the accuracy of the rating similarity.In order to verify the effectiveness and feasibility of the proposed method, we use Movie Lens dataset in the nearest neighbor query efficient and the recommender precision to analyze and compare operations. The experimental results show that the project-based hierarchical clustering cohesion collaborative filtering recommendation algorithm the nearest neighbor query efficient and the recommender precision are better than other traditional collaborative filtering algorithms.