Stability Analysis for Users´ Web Preference

Advanced personalized e-applications require comprehensive knowledge about their user´s preference in order to provide individual product recommendations and custom-tailored product offers. Much research has been conducted using Web logs to infer user preferences and predict users´ behavior. However, little research measures the stability of user preferences over time. In this paper, the index named Web stability coefficient (ST) based on a comprehensive analysis of Web access records is proposed to represent the stability of user´s preference. Then we used k-means clustering algorithm to assign users into different groups, which is computed based on the values of vector representation of user´s ST. By analyzing user patterns, we present some interesting conclusions that facilitate us to better understand behavioral characteristics of Web user.