Distributed RDF query processing and reasoning for Big Data Linked Data

The Linked Data Movement is aimed at converting unstructured and semi-structured data on the documents to semantically connected documents called the "web of data." This is based on Resource Description Framework (RDF) that represents the semantic data and a collection of such statements shapes an RDF graph. SPARQL is a query language designed specifically to query RDF data. Linked Data faces the same challenge that Big Data does. We now lead the way to a new wave of a new paradigm, Big Data and Linked Data that identify massive amounts of data in a connected form. Indeed, utilizing Linked Data and Big Data continue to be in high demand. Therefore, we need a scalable and accessible query system for the reusability and availability of existing web data. However, existing SPAQL query systems are not sufficiently scalable for Big Data and Linked Data.;In this thesis, we address an issue of how to improve the scalability and performance of query processing with Big Data / Linked Data. Our aim is to evaluate and assess presently available SPARQL query engines and develop an effective model to query RDF data that should be scalable with reasoning capabilities. We designed an efficient and distributed SPARQL engine using MapReduce (parallel and distributed processing for large data sets on a cluster) and the Apache Cassandra database (scalable and highly available peer to peer distributed database system). We evaluated an existing in-memory based ARQ engine provided by Jena framework and found that it cannot handle large datasets, as it only works based on the in-memory feature of the system. It was shown that the proposed model had powerful reasoning capabilities and dealt efficiently with big datasets.