Collaborative Filtering Algorithm Oriented User Interests

Personalized recommendation is the main technology that solves the problem of information overload.Collaborative filtering is one of the most successfully algorithms;it analyzes user's interest bases on the user's log records to finds the nearest neighbor and recommend item sets.There are many problems in the application because user's interest is dynamic.To solve the problems of user interest drift,sparseness and accuracy in the traditional collaborative filtering algorithm,the thesis studies from the following aspects:(1)The thesis overviews the home and abroad status about the user interest drift and timeliness,and analyzes the relevant theories.(2)To improve the accuracy and alleviate the user interest drift,a new collaborative filtering algorithm is proposed based on user interest and item properties.Firstly the user interest weight is added to the process of similarity calculating,and then integrated with the item properties similarity.(3)An improved similarity calculation method based on user-based collaborative filtering algorithm is presented by using Bhattacharyya coefficient.In the process of the similarity computing the user's interest weight is added.The experimental results on the MovieLens dataset demonstrate that our approach has advantages over state-of-the-art methods in terms of both the discovery of user interest preference and providing highly accuracy recommendations.According the theory of this thesis designed the movie prototype system,to achieve the expected requirements of the project.