A Personalized Recommendation Algorithm Based on Comprehensive Interest

摘要

personalized recommender systems have been widely used in E-commerce to help customers find interesting products more conveniently. However, with the development of E-commerce, the magnitudes of users and commodities grow rapidly. Traditional recommendation algorithms face severe challenge of sparse user rating and real-time recommendation. To address these issues, the user's behavior sequence and the merchandise category structure are analyzed deeply and a personalized recommendation algorithm based on comprehensive interest is proposed. According to the behaviors information of browsing, purchasing and rating in an actual E-commerce process, the user-item rating matrix has been changed to user-item comprehensive interest matrix. It can not only reflect the real interest of users, but also alleviate the sparse of the rating matrix to large extent. Furthermore, a user has different interest on different category. The searching of nearest neighbors and the predicting of interest for a certain item should be within the same category. Therefore, the searching space can be reduced and the accuracy can be improved. Simulation results on three data sets of a real E-commerce system illustrate the efficiency of this algorithm.