Web behaviormetric user profiling concept

Abstract

We present a concept for building behaviorally centered user profiles. The concept utilizes behavioral analytics of user interactions in web environments. User interactions are temporally segmented into elemental browsing units. The browsing segments permit identification of the essential navigational points as well as higher order abstractions. The profiles incorporate relevant metrics from three major domains: temporal, navigational, and abstractions. Temporal metrics focus on aspects of durations and delays between portions of human interactions. The navigational metrics target the initial, terminal, and single user actions. The abstraction metrics encompass elemental patterns of human browsing behavior and their interconnections. The profiling concept utilizes relatively simple analytic and statistical apparatus. It facilitates computational efficiency and scalability to large user domains. © Springer-Verlag Berlin Heidelberg 2008.