The RDF data model is a key technology in the Linked Data vision. Given its graph structure, even relatively simple RDF queries often involve a large number of joins. Join evaluation poses a significant performance challenge on all state-of-the-art RDF engines. TripleT is a novel RDF in- dex data structure, demonstrated to be competitive with the current state-of-the-art for join processing. Query opti- mization on TripleT, however, has not been systematically studied up to this point. In this paper we investigate how the use of (i) heuristics and (ii) data statistics can contribute to- wards a more intelligent way of generating query plans over TripleT-based RDF stores. We propose a generic framework for query optimization, and show through an extensive em- pirical study that our framework consistently produces efi- cient query evaluation plans. © 2015, Copyright is with the authors.
Keywords: Indexing; Query processing; RDF; SPARQL; TripleT
|Name||CEUR Workshop Proceedings|
|Conference||conference; EDBT/ICDT 2015 Joint Conference; 2015-03-27; 2015-03-27|
|Period||27/03/15 → 27/03/15|
|Other||EDBT/ICDT 2015 Joint Conference|