A method of using string-matching to analyze hypertext navigation was developed, and evaluated using two weeks of website logfile data. The method is divided into phases that use: (i) exact string-matching to calculate subsequences of links that were repeated in different navigation sessions (common trails through the website), and then (ii) inexact matching to find other similar sessions (a community of users with a similar interest). The evaluation showed how subsequences could be used to understand the information pathways users chose to follow within a website, and that exact and inexact matching provided complementary ways of identifying information that may have been of interest to a whole community of users, but which was only found by a minority. This illustrates how string-matching could be used to improve the structure of hypertext collections. Categories and Subject Descriptors H.3.3 [Information Storage and Retrieval]: Information Search and retrieval – clustering, searc...
Roy A. Ruddle