An Implementation of Electronic Commerce Recommender System Based on Improved K-means Clustering Algorithm

Recommender system plays an important role in modern Electronic Commerce. An excellent recommender system is the key to make electronic commerce network run well. But now because of many reasons most recommend effects are not good enough. So a recommender system based on Improved K-means Clustering Algorithm (IKCA) is designed and implemented in this paper. The whole system includes user clustering module, prediction recommending module and evaluating module. This paper also studies and analyzes the influence factors of recommend effect and improves recommending accuracy. Traditional K-means Clustering Algorithm (TKCA) often falls into local optimal solution. IKCA uses the moving operator to adjust distance from user to cluster centre so it can more easily escape from local optimal solution and approach the global optimal. The experimental result shows that IKCA is better than TKCA. This system can be generally applied in the other fields.