Research on path clustering based on the access interest of users

Abstract

Users with same interests can be classified by making use of clustering technology, based on the access path of users, the page access time of users, the resident time at page and URL of linking to the page in a web site. A new clustering algorithm is proposed by using access interests of users in the paper, in which the new interest degree, similitude degree, and clustering center will be defined. A true experiment has been completed making use of log files in the www.ty.sx.cn web site. Experiment results are successful. © Springer-Verlag Berlin Heidelberg 2005.