Towards Mining Semantic Maturity in Social Bookmarking Systems.
Martin Atzmueller, Dominik C. Benz, Andreas Hotho, Gerd Stumme · 2011
Abstract. The existence of emergent semantics within social metadata (such as tags in bookmarking systems) has been proven by a large number of successful approaches making the implicit semantic structures explicit. However, much less attention has been given to the factors which influence the “maturing ” process of these structures over time. A natural hypothesis is that tags become semantically more and more mature whenever many users use them in the same contexts. This would allow to describe a tag by a specific and informative “semantic fingerprint ” in the context of tagged resoures. However, the question of assessing the quality of such fingerprints has been seldomly addressed. In this paper, we provide a systematic approach of mining semantic maturity profiles within folksonomy-based tag properties. Our ultimate goal is to provide a characterization of “mature tags”. Additionally, we consider semantic information about the tags as a gold-standard source for the characterization of the collected results. Our initial results suggest that a suitable composition of tag properties allows the identification of more mature tag subsets. The presented work has implications for a number of problems related to social tagging systems, including tag ranking, tag recommendation, and the capturing of light-weight ontologies from tagging data. 1