SEMANTIC GROUNDING STRATEGIES FOR TAGBASED RECOMMENDER SYSTEMS

Recommender systems usually operate on similarities between recommended items or users. Tag basedrecommender systems utilize similarities on tags. The tags are however mostly free user entered phrases.Therefore, similarities computed without their semantic groundings might lead to less relevantrecommendations. In this paper, we study a semantic grounding used for tag similarity calculus. We show acomprehensive analysis of semantic grounding given by 20 ontologies from different domains. The studybesides other things reveals that currently available OWL ontologies are very narrow and the percentageof the similarity expansions is rather small. WordNet scores slightly better as it is broader but not much asit does not support several semantic relationships. Furthermore, the study reveals that even with suchnumber of expansions, the recommendations change considerably