Semantic relatedness in DBpedia: A comparative and experimental assessment

Evaluating semantic relatedness of Web resources is still an open challenge. This paper focuses on knowledge-based methods, which represent alternative to corpus-based approaches, and rely in general the availability knowledge graphs. In particular, we have selected 10 methods from existing literature, that been organized according it adjacent resources, triple patterns, weights-based methods. They implemented evaluated by using DBpedia as reference RDF graph. Since continuously evolving, experimental results provided these literature are not comparable. For this reason, work, such experimented running them all at once same release against 14 well-known golden datasets. On basis correlation values with human judgment obtained results, weighting triples combination evaluating directed paths linking compared best strategy order compute DBpedia.