Explore the Structure of Social Tags by Subsumption Relations

摘要

Thanks to its simplicity, social tagging system has accumulated huge amount of user contributed tags. However, user contributed tags lack explicit hierarchical structure, while many tag-based applications would benefit if such a structure presents. In this work, we explore the structure of tags with a directed and easy-to-evaluate relation, named as the subsumption relation. We propose three methods to discover the subsumption relation between tags. Specifically, the tagged document’s content is used to find the relations, which leads to better result. Besides relation discovery, we also propose a greedy algorithm to eliminate the redundant relations by constructing a Layered Directed Acyclic Graph (Layered- DAG) of tags. We perform quantitative evaluations on two real world data sets. The results show that our methods outperform hierarchical clustering-based approach. Empirical study of the constructed Layered-DAG and error analysis are also provided.