Comparing DBpedia, Wikidata, and YAGO for Web Information Retrieval

Knowledge graphs serve as the primary sources of structured data in many SemanticWeb applications. In this paper, the three most popular cross-domain knowledge graphs (KGs), namely, DBpedia, YAGO, andWikidata were empirically explored and compared. These knowledge graphs were compared from the perspectives of completeness of the relations, timeliness of the data and accessibility of the KG. Three fundamental categories of named entities were queried within the KGs for detailed analysis of the data returned. From the experimental results and findings, Wikidata scores the highest in term of the timeliness of the data provided owing to the effort of global community update, with DBpedia LIVE being the next. Regarding accessibility, it was observed that DBpedia andWikidata gave continuous access using public SPARQL endpoint, while YAGO endpoints were intermittently inaccessible. With respect to completeness of predicates, none of the KGs have a remarkable lead for any of the selected categories. From the analysis, it is observed that none of the KG can be considered complete on its own with regard to the relations of an entity.