An Apparel Recommender System Based on Data Mining

The rapid e-commerce growth has made both business community and customers face a new situation. Web usage mining attempts to discover useful knowledge from the secondary data obtained from the interactions of the customers with the Web. In this paper, we present a case study of an on-line system that recommends apparels based on a knowledge base that consists of rules gotten from decision tree mining and experienced dressing knowledge. To get these rules, apparel components features such as collar style, number of buttons, slits, kind of fabrics, color and apparel style, etc. that characterize the domain of interest in our case must be extracted at first. Then the components features of apparel products that customers browse or purchase are analyzed and a decision tree model which could infer the customers' "tastes" from their personal information could be gotten. Therefore, the system could recommend apparel items catering customers "tastes" according the decision tree model. The architecture and general technology of our on-line recommender system are described also. We believe that this approach is relevant to a wider class of e-commerce problems and it can be used in a variety of recommender systems.