Using incremental fuzzy clustering to web usage mining

Abstract

The recent extensive growth of data on the Web, has generated an enormous amount of log records on Web server databases. Applying Web Usage Mining techniques on these vast amounts of historical data can discover potentially useful patterns and reveal user access behaviors on the Web site. Cluster analysis has widely been applied to generate user behavior models on Server Web logs. Most of these off-line models have the problem of the decrease of accuracy over time resulted of new users joining or changes of behavior for existing users in modelbased approaches. This paper proposes a novel approach to generate dynamic model from off-line model created by fussy clustering. In this method, we will use users' transactions periodically to change the off-line model. To this aim, an improved model of leader clustering along with a static approach is used to regenerate clusters in an incremental fashion. © 2009 IEEE.