JOTR: Join-Optimistic Triple Reordering Approach for SPARQL Query Optimization on Big RDF Data

Resource Description Framework (RDF) is increasingly being used for representing information on the web. This popularity has made storage of large RDF data a difficult task. To overcome these issues many distributed RDF systems are being proposed that can store and efficiently process Big RDF data. Hadoop framework is widely being used for storing and handling a large amount of RDF data. One of the major obstacles faced while handling this large amount of RDF data is query processing on such large datasets. In this paper, we present JOTR: a SPARQL query optimization technique for Big RDF data using triple pattern reordering on a distributed Hadoop based RDF system. The proposed technique is based on selectivity calculation and has been tested on one of the popular RDF benchmark datasets, LUBM dataset. We have tested JOTR on large sized RDF datasets and compared it with other optimization approaches in respect to the query execution time. From the results, it can be concluded that our approach gives a notable performance on distributed RDF systems and thus is applicable to centralized systems as well.