Automatically identifying tag types

Abstract

Web 2.0 applications such as delicious, flickr or lastfm have recently become extremely popular and as a result, a large amount of semantically rich metadata produced by users becomes available and exploitable. Tag information can be used for many purposes (e.g. user profiling, recommendations, clustering etc), though the benefit of tags for search is by far the most discussed usage. Tag types differ largely across systems and previous studies showed that, while some tag type categories might be useful for some particular users when searching, they may not bring any benefit to others. The present paper proposes an approach which utilizes rule-based as well as model-based methods, in order to automatically identify exactly these different types of tags. We compare the automatic tag classification produced by our algorithms against a ground truth data set, consisting of manual tag type assignments produced by human raters. Experimental results show that our methods can identify tag types with high accuracy, thus enabling further improvement of systems making use of social tags. © 2009 Springer.