Research on personalized recommendation algorithm based on user model and user-project matrix

With the dramatic increase of the amount of network information, the phenomenon of merchandise information overload has become more and more serious. Thus, the e-commerce sites are required to provide the most needed information for customers to attract their attention. The traditional recommendation algorithm ignores the connections among user´s own attributes and the changes in project scoring with each passing day. With the research of the traditional collaborative filtering recommendation algorithm, the paper put an emphasis on the combination of user model and user project matrix, and introduced the data of user attributes and the variation values of project scoring over time into the traditional recommendation algorithm. By experiment, the validity of the algorithm was confirmed.