Towards a Web of Semantic Tags

Folksonomies are widely used to classify content on the Web. However, plain text annotations hardly fit the vision of the Semantics Web, where an unambiguous understanding of the data by both user and machine is key. Ontologies underpinning the Web help to resolve the ambiguity problem, but a.o. the absence of a detailed world ontology still puts end-user tagging in the front seat for basic content classification. In this paper, we propose semantic tags to bridge the gap between plain text keywords and ontologies. Sematags define aliases (synonyms) and isas (hypernyms) to better cope with the issues traditional tags suffer from. We illustrate how sematags can be defined from scratch or extracted from existing lexicons and knowledge bases. To evaluate our approach, we composed sematags from Wikipedia concepts and used those to semiautomatically tag photos on Flickr.