Structural Properties as Proxy for Semantic Relevance in RDF Graph Sampling

摘要: The Linked Data cloud has grown to become the largest knowledge base ever constructed. Its size is now turning into a major bottleneck for many applications. In order facilitate access this structured information, paper proposes an automatic sampling method targeted at maximizing answer coverage applications using SPARQL querying. approach presented in novel: no similar RDF exist. Additionally, concept of creating sample aimed coverage, unique. We empirically show that relevance triples (a semantic notion) influenced by topology graph (purely structural), and can be determined without prior queries. Experiments significantly higher recall based methods over random naive baseline approaches (e.g. up 90% Open-BioMed 6%).