Combined bit map representation and its applications to query processing of resource description framework on GPU. (8th November 2019)

• Record Type: Journal Article
• Authors: Chantrapornchai, Chantana; Choksuchat, Chidchanok
• Is Part Of: International journal of high performance computing and networking. Volume 15:Number 1/2(2019)
• Page Start: 45
• Page End: 57
• Publication Date: 2019-11-08
• Subjects: graphics processing unit -- GPUs -- semantic web -- query processing -- parallel processing -- bit map; High performance computing -- Periodicals; Computer networks -- Periodicals; High performance computing; Periodicals; 004.05
• Languages: English
• ISSNs: 1740-0562

Abstract

Resource description framework (RDF) is a common representation in semantic web context, including the web data sources and their relations in the URI form. With the growth of data accessible on the Internet, the RDF data currently contains millions of relations. Thus, answering a semantic query requires going through large amounts of data relations, which is time consuming. In this work, we present a representation framework, combined bit map representation (CBM), which compactly represents RDF data while helping speed up semantic query processing using graphics processing units (GPUs). Since GPUs have limited memory size, without compaction the RDF data cannot be entirely stored in the GPU memory; the CBM structure enables more RDF data to reside in the GPU memory. Since GPUs have many processing elements, their parallel use speeds up RDF query processing. The experimental results show that the proposed representation can reduce the size of RDF data by 70%. Furthermore, the search time on this representation using the GPU is 60% faster than with conventional implementation.