Paper Info
Title
Backward inference and pruning for RDF change detection using RDBMS.
Abstract
Recent studies on change detection for RDF data have focused on minimizing the delta size and, as a way to exploit the semantics of RDF models in reducing the delta size, the forward-chaining inferences have been widely employed. However, since the forward-chaining inferences should pre-compute the entire closure of the RDF model, the existing approaches are not scalable to large RDF data sets. In this paper, we propose a scalable change detection scheme for RDF data, which is based on backward-chaining inference and pruning. Our scheme, instead of pre-computing the full closure, computes only the necessary closure on the fly, thus achieving fast and scalable change detection. In addition, for any two RDF data input files to be compared, the delta obtained from our scheme is always equivalent to the one from the existing forward-chaining inferences. In addition, in order to handle RDF data sets too large to fit in the available RAM, we present an SQL-based implementation of our scheme. Our experimental results show that our scheme, in comparison to the existing schemes, can reduce the number of inference triples for RDF change detection by 10-60%.
Year
DOI
Venue
2013
10.1177/0165551512463650
JOURNAL OF INFORMATION SCIENCE
Keywords
DocType
Volume
backward-chaining inference,change detection,ontology,pruning,RDF
Journal
39.0
Issue
ISSN
Citations 
2
0165-5515
0
PageRank 
References 
Authors
0.34
0
3
Name
Order
Citations
PageRank
Dong-Hyuk Im1356.06
Sang-Won Lee21536106.03
Hyoung-Joo Kim31108374.05