E-commerce web site contains a large user access to information., by data mining, can enhance the site to an accurate understanding of user access to information to enhance e-commerce web site click-through rate. To do this will extract the e-commerce website logs the user access to link data, the use of denoising techniques logging user access to link data filtering analysis, and the data was became binary array by dissimilarity binary relation, so this data will be computed by differentiation analysis, and the realization electricity commerce website user，s cluster, will provide the reference for the website page optimization and user’s interest and habbies.