Complex Networks of Human WEB Behavior
Last modified: August 22, 2007
Understanding human behavior in electronic environments is vital for designing intelligent personalized technologies deployable at internet and intranets. The problem of human web behavior analysis has been attracting substantial attention from commercial sectors as well as academic establishments. Internet and software vendors have been going to extreme extents in behavioral data acquisition. Web log data, however, has been largely underutilized for behavioral analysis. The data is generally voluminous, incomplete, and contaminated; which make the analysis particularly difficult. We introduce a novel analytic framework that efficiently captures the spatiotemporal dimensions of human dynamics in electronic spaces. It enables elucidation of browsing behavior at both elemental and higher order abstraction levels. The framework has been utilized in analyzing knowledge worker behavior on a large intranet. The analysis revealed numerous important findings. Knowledge workers exhibited significant tendency to form elemental and complex browsing patterns. They had focused interests and displayed diminutive exploratory behavior. Topology of their web traversal pathways corresponds to the complex networks at both the elemental and higher order abstraction levels. Human browsing network topology, however, does not match the structural topology of the web environment. These novel discoveries have significant impact in scientific and commercial domains. They clarify why the conventional personalization approaches are ineffective, and highlight potential avenues for new advancements.