A Novel Approach for Clustering Web User Sessions Using RST

Abstract

Web usage mining has assumed importance in learning about web user's behavior and user interactions with the website. It uses data mining techniques to discover non-trivial user behavior patterns. These patterns can then be used to make the predictions of next page to be accessed by the user. Web usage mining consists of the steps like web log preprocessing, pattern discovery and pattern analysis. This paper proposes a novel approach for preprocessing wherein rough set clustering is applied to form the clusters of sessions. These sessions could later on be used to form the knowledge base of rules on the basis of which the next page to be accessed could be prefetched.

Session identification techniques used in web usage mining

“…Sessionization heuristics can be broadly classified as: time-oriented (Zhou et al , 2006; Castellano et al , 2007; Peng and Zhao, 2010; Xinhua and Qiong, 2011; Dinuca and Ciobanu, 2012) and structure-oriented (Khasawneh and Chan, 2006; Fang and Huang, 2010). One approach for session identification is to use clustering techniques for grouping user sessions (Murray et al , 2006; Alam et al , 2008; Lu and Nguyen, 2009; Jyoti et al , 2009; Hussain et al , 2010; Chordia and Adhiya, 2011; Huidrom and Bagoria, 2013; Chang-bin and Li, 2010). Later, link-based sessionization heuristics were proposed as an alternate for time-oriented and structure-oriented heuristics (Bayir et al , 2009; Bayir and Toroslu, 2013; Tiknaz, 2013).…”

“…In order to determine the interesting user patterns, Jyoti et al (2009) presented a WUM scheme to identify individual user navigational behavior on the web. Experimental results showed that the algorithm outperforms the traditional clustering techniques in order to derive groups of users which exhibit similar access patterns.…”

“…In order to determine the interesting user patterns, Jyoti et al (2009) presented another WUM scheme to identify individual user navigational behavior on the web. These patterns can then be used to predict the next sequence of pages accessed by the user.…”

Semantically Enriched Keyword Prefetching Based on Usage and Domain Knowledge

“…There was a test cases series that was executed for a variety of sessions, ranging from shorter ones to longer ones. For the WUM method in our trials, we employed the Markov model-based technique [25] and the association rule mining technique [26]. For the WCM technique, we used the keyword-based approach [26].…”