[2H1-GS-3a] 知識の利用：基礎

[2H1-GS-3a-03] Efficient Index Structures for Node Search of RDF Graphs

〇Koji Fujiwara¹, Ken Kaneiwa¹ (1. Graduate School of Informatics and Engineering, The University of Electro-Communications)

Keywords:Index Structures, RDF Graphs, Node Search, Data Compression, SPARQL

Published RDF data on the Web have been increasing, and it is important to execute data mining and machine
learning on RDF graphs. Most RDF stores build index structures for efficient storages and retrieval of huge RDF
graphs. However, it is inefficient because multiple indexes are required to match any triple patterns, and duplicate
results in neighbor node search for data mining and machine learning feature extraction. In this work, we propose
index structures called a P-Link index for fast query resolving and efficient data storing.The P-link index focuses
on the fact that the subject and object of a triple are nodes of an RDF graph, it has an index structure for node
search, and predicate links for matching any triple pattern. Our experiments show that the P-Link index with an
index structure reduces the memory space and efficiently executes specific search tasks such as node search.