Semantic Query Processing Over Linked Data Knowledge Bases

Query processing is an important feature and component of any information management system. On the Semantic Web, intelligent query processing is mostly the difference in rendering an underlying RDF data store valuable and useful for applications. Query processing systems are expected to efficiently process queries over datastores without any degradation in performance, even when the data size increases.The potential of live query processing to discover new information from disparate datasources has made it an increasingly popular query processing strategy for linked data. Different implementations and combinations with other query processing strategies have been proposed and implemented by researchers in the linked data research community. In this thesis,1) we propose a query processing model that uses a live query processing strategy to leverage the linked data web to serve as an additional source of information for linked data applications.2) We also present an algorithm (called Seed Populator Algorithm) that can be used to provide additional relevant datasource identifiers as seed URIs to the query execution system. A very important contribution of this Seed Populator algorithm is that it helps to solve the problem of Unconnected Query-Relevant Documents highlighted by Umbrich. A problem that limits the amount of query results retrieved by an execution engine because query-relevant documents cannot be reached due to unavailability of reference links in already retrieved results.3) We further describe and present a general algorithm used by an execution system to directly query the linked data web. This algorithm is called the Live Query Execution Algorithm. In our model and experimentation, we show that the more relevant datasources are made available to a query execution system, the more likely for the query system to discover more potential results.4) We also present a sample linked data application called Research Information Discovery that uses a live query execution system to leverage the linked data web as an additional source of information.The significance of the linked data web is to provide a globally distributed and interconnected database of useful knowledge from all works of life. Through efficient query processing, applications have the potential of discovering and utilizing this relevant information. In this thesis, we show how linked data applications can leverage such enormous amounts of information from the linked data web.