[4M2-GS-5] Web intelligence: Web intelligence

[4M2-GS-5-04] Umaka Search: A search service for SPARQL endpoint

〇Yasunori Yamamoto¹, Atsuko Yamaguchi¹ (1. Database Center for Life Science)

Keywords:Semantic Web, Linked Data, SPARQL Endpoint

We have developed a SPARQL endpoint search system Umaka Search, which accepts any URIs and returns relevant SPARQL endpoints. Relevant here means that these SPARQL endpoints have RDF datasets where the given URI should appear as subjects. As the sizes of RDF datasets in life sciences are often a couple of billion triples, these indexes also become large and consume machine resources that are too costly to provide this service. We proposed a way of reducing the index sizes by removing insignificant parts of URIs to search for SPARQL endpoints. As a result, the sizes have been reduced to fit an ordinary machine environment at the cost of losing the ability to check existence of every URI in the target SPARQL endpoints.