Joint Item-Tag Recommendation Framework for Collaborative Filtering in Social Tagging Systems

Abstract

Tapping into the wisdom of the crowd, social tagging is becoming an increasingly important mechanism for organizing and discovering information on the Web. Effective tag-based recommendation of information items is one of the key technologies contributing to the success of this social information discovery mechanism. A precise understanding of the information structure of social tagging systems lies at the core of an effective tag-based item recommendation method. While most existing methods either implicitly or explicitly assume a simple tripartite graph structure, in this paper, we propose a comprehensive data model to capture all types of co-occurrence information in the social tagging context. Based on this data model, we further propose a unified user profiling scheme to make full use of all available information. Finally, supported by this user profile, we propose a framework for collaborative filtering in social tagging systems. In this framework, we first generate joint item-tag recommendations, with tags indicating topical interests of users in target items. These joint recommendations are then refined by the wisdom from the crowd and projected to the item (or tag) space for final item (or tag) recommendations. Empirical evaluation using real-world data demonstrates the utility of our proposed approach.