Semantic Grounding Strategies for Tagbased Recommender Systems

TL;DR

This paper investigates semantic grounding methods for tag-based recommender systems, analyzing how different ontologies influence tag similarity and recommendation relevance, revealing limitations of current ontologies and the impact of semantic expansions.

Contribution

It provides a comprehensive analysis of semantic grounding using 20 ontologies, highlighting the limitations of OWL ontologies and WordNet in enhancing tag similarity for recommendations.

Findings

OWL ontologies are narrow with limited similarity expansions

WordNet offers broader coverage but lacks certain semantic relationships

Semantic expansions significantly alter recommendation outcomes

Abstract

Recommender systems usually operate on similarities between recommended items or users. Tag based recommender 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 relevant recommendations. In this paper, we study a semantic grounding used for tag similarity calculus. We show a comprehensive analysis of semantic grounding given by 20 ontologies from different domains. The study besides other things reveals that currently available OWL ontologies are very narrow and the percentage of the similarity expansions is rather small. WordNet scores slightly better as it is broader but not much as it does not support several semantic relationships. Furthermore, the study reveals that even with such number of expansions, the recommendations change considerably.