Research And Application Of Recommendation Technology Based On User's Interest

Personalized recommendation system,as an intelligent recommendation platform,is well convenient for users to retrieve information.Collaborative filtering algorithm is the core technology.As the data size grows up rapidly,the problem of sparsity of data is becoming more and more serious.As a result,traditional collaborative filtering recommendation technology is facing with a lot of problems,such as difficulty of user interest modeling,low accuracy rate of recommendation results and so on.In order to solve the above problem,this paper makes an in-depth analysis and researchThis paper proposes a new user interest modeling algorithm based on entropy method.This algorithm firstly sumarizes the user behaviors according to the item attributes and then uses the entropy method to set weight of the behavior indexes objectively and further evaluates user interests comprehensively.The experimental results show that this algorithm can exploit the users' interest effectively,even if the sparsity of data problem is very serious,so as to improve the performance of the recommendation system.This paper also proposes a hierarchical collaborative filtering recommendation algorithm based on user reliability.This algorithm firstly selects the users who have similar interests with the target user to be the candidate neighbors,and then selects the final neighbors from the candidate neighbors.It adds a reliability factor of the neighbors on the basis of the traditional collaborative filtering algorithm Comparative results demonstrate that the proposed algorithm can greatly improve the accuracy of recommendation results.The proposed algorithm is applied in the field of telecommunication and the results from the real data show that the algorithm has excellent generalization ability.