Tagpref: User Preference Modeling by Social Tagging

User preference modeling is the basis for the provision of personal information services because it can reflect the user´s characteristics and interests. Tagging is a popular online activity that allows a user to discover, describe, and organize Internet contents. Tagging provides a new means of creating user preference. Tags explain and describe items so that a user´s attitude toward a tag is his or her attitude toward an aspect of the item. In this paper we design Tagpref inspired by the method of establishing user preference with tags. The user´s preference was described by the user´s preference degrees for different tags. Different algorithms were proposed and evaluated for taggers and non-taggers on Movie Lens based on direct and indirect behavior. Evaluation results show that the proposed algorithms can accurately predict a user´s tag preference. The user´s attitude toward tag preference was also analyzed. Users´ true preference was determined by an online user survey.