A personalized recommendation technology for E-commerce website
Hongqing Guo, Li Chen, Sanxin Chen, Jian Mi · 2016
This paper presents a new personalized recommendation technology for e-commerce Web site, which combines clustering users' expectations and Item-Based Collaborative Filtering recommendation algorithm. Similar distance between two websites means similar expectations. Firstly we cluster the Web sites by calculating the distance between any two Web sites. Naturally, the expected distance matrix of Web sites cluster is attained. Then according to the distance matrix, the expected similarity matrix can be calculated, and the nearest neighbor Web site set can be searched. Current users calculate their prediction scores for every Web site in the nearest neighbor set. The Web sites with top N scores are recommendations. This paper only searches in the space produced by clustering, which improves the efficiency. It starts from the demand of users and the Web sites not visited by a majority may be recommended. So the method effectively punishes the case of “the most popular Web site”.