LODS: A Linked Open Data Based Similarity Measure

Abstract

With the rapid evolution of Linked Open Data (LOD), researchers are exploiting it to solve particular problems such as semantic similarity assessment. Existing LOD-based semantic similarity approaches attach compared data (terms or concepts) to LOD resources to exploit their semantic descriptions and relationships with other resources and estimate the degree of overlap between resources. Current approaches suffer from two limitations: they focus on the analysis of links between resources and ignore the important taxonomic structure of concepts and categories used to describe resources. On the other hand, they do not exploit interlinks between LOD resources in order to enrich data used to compute the similarity score. In this paper, we overcome the above limitations by proposing a new LOD-based similarity measure based on the combination of ontological, classification and property dimensions of LOD resources.