GADES: A graph-based semantic similarity measure

Knowledge graphs encode semantics that describes resources in terms of several aspects, e.g., neighbors, class hierarchies, or node degrees. Assessing relatedness of knowledge graph entities is crucial for several data-driven tasks, e.g., ranking, clustering, or link discovery. However, existing similarity measures consider aspects in isolation when determining entity relatedness. We address the problem of similarity assessment between knowledge graph entities, and devise GADES. GADES relies on aspect similarities and computes a similarity measure as the combination of these similarity values. We empirically evaluate the accuracy of GADES on knowledge graphs from different domains, e.g., proteins, and news. Experiment results indicate that GADES exhibits higher correlation with gold standards than studied existing approaches. Thus, these results suggest that similarity measures should not consider aspects in isolation, but combinations of them to precisely determine relatedness.