Customer Behavior Pattern Discovering with Web Mining

This paper describes a real application proven web mining approach. The approach performs with integrated data comprised of web logs and customer information involved in e-commerce web sites. The objective is to acquire behavior patterns of visitors on web sites. The mining tasks include the customer clustering, association rules among the web pages of visitor traffic, buying patterns of customers, and predict model generation for the potential customers. As web log data is very extraneous, low granularity and voluminous, a semantic taxonomy method is used to group web pages, helping address the discovered patterns. This web mining work is useful for an enterprise to have a multi-level customer view, which prompts decision-making process of the enterprise.