User Session Identification Using Enhanced Href Method

Jozef Kapusta, Peter Švec, Michal Munk, Ján Skalka

Abstract


One part of web log mining covers the process of discovering web users’ behavior patterns. This process employs user session identification methods based on the web structure. There are some common heuristic methods which use referral URL from the web server log file. Using the referral URL is an alternative technique for the user session identification. We identified possible deficiencies of the common used h-ref methods and propose its improvements. We applied the improved h-ref method on the web server log file and thus identify user session in a different way. Next we compared basic characteristics of extracted user behavior rules using the descriptive statistic methods among different h-ref methods. Results of the experiment show that the improved h-ref method does not affect session identification, only affect the inclusion of page visits into existing sessions. The new h-ref method is as effective as the generic one.


Back to list of accepted papers