The Research Of Collaborative Filtering Recommendation Algorithm Based On User Interest Change

The rapid development of the Information Technology has made more and more web users take part in the creating and sharing of the Internet information. The web users were not only browsing but also sending and sharing the resources or information,which has made the Information Overload become a big and unavoidable problem. In this situation, the personalized recommender system appeared. While with the scale of the resources and users becoming bigger and bigger, many problems of recommender system occurred to people. This paper have had a deep research and analysis on the collaborative filtering technology which is applied the most widely in the recommendation. Also, this paper proposed an improved algorithm for the problem of the collaborative filtering technology.Firstly, this paper described the basic knowledge of personalized recommender system and analyzed the main recommendation technology. Especially for the collaborative filtering which is the most mature and widely used, this paper listed some problems and challenges.Secondly, enlightened by Ebbinghaus curve, time factor was added in the traditional Collaborative Filtering Recommendation Algorithm to solve the problem of people interest change. Meanwhile, an improved algorithm based on forgetting function was proposed.Finally, this paper compared and analyzed the improved algorithm based on forgetting function with the traditional collaborative filtering algorithm. Then the final conclusion was made that the improved algorithm based on use interest change is better than the traditional collaborative filtering algorithm.