Skip to Main Content
When users access a Web site, the access of the users represents the interest of the users in the Web pages of the Web site. Each user's preference can be manifested by the sequence of access sequence. After identifying each user's access transactions and defining a similarity measure between two transactions, the access paths of all the users can be clustered. This is called path clustering. In this paper, we defined a new similarity measure between two web user's transactions. It simultaneously takes the page co-occurrence frequencies and local sequence of access path into consideration. Based on this measurement we presented an algorithm of path clustering. The simulated experimental results show that the algorithm maybe more applicable in some situation.