Authors: Jongmo Kim; Kunyoung Kim; Mye Sohn; Gyudong Park
Addresses: Department of Industrial Engineering, Sungkyunkwan University, Suwon, South Korea ' Department of Industrial Engineering, Sungkyunkwan University, Suwon, South Korea ' Department of Industrial Engineering, Sungkyunkwan University, Suwon, South Korea ' 2nd R&D Institute, Agency for Defense Development, Seoul, South Korea
Abstract: Linked open data (LOD) has emerged as a new platform for sharing and integrating information about not only web resources but also physical resources. However, as the volume of LOD increases explosively, it becomes difficult to query the LOD to discover the high-quality information. To obtain the desired information from the LOD, we propose the query graph extension framework (Q-PD framework) that can extend the users' queries. To do so, the Q-PD framework identifies the entity graph (EG), which contain the LOD resources directly related to the users' queries. As a next, it performs predicate-based RDF clustering to find the topic graph patterns (TGPs), which are the EGs to be reconstituted with information on the specific topics. Finally, the Q-PD framework extends the RDF graph patterns related to the users' queries using the TGPs. To prove the excellence of the Q-PD framework, we performed the three kinds of experiment with 681.9 entities and 51,577.1 RDF triples collected from DBpedia. Experimental results show that the Q-PD framework is superior to the existing bottom-up approach in terms of completeness and accuracy.
Keywords: semantic web; linked open data; LOD; graph clustering; graph pattern recognition; SPARQL query rewriting.
International Journal of Web and Grid Services, 2020 Vol.16 No.2, pp.105 - 125
Received: 02 Aug 2019
Accepted: 17 Jan 2020
Published online: 23 Jun 2020 *