With the advent of pervasive computing, particularly in a resource-limited environment where electrical devices such as alarm clocks, coffee makers, vehicles, and refrigerators search and reason over data generated by their operation, it is important that data is described in a form suitable for the machines to interpret and understand in a consistent manner. Semantic Web technologies could play a key role in bringing about the vision of a semantic-oriented pervasive computing environment. Nowadays various computer systems and applications use Semantic Web languages such as Web Ontology Language (OWL) as a foundation for representing and sharing information data of interest.
Automatic reasoning is the major benefit of using the data, i.e., Semantic Web ontologies, expressed by Semantic Web languages. The reasoning is a process of drawing an inference from the asserted information (i.e., axioms) in an ontology based on the well-known logical procedures, and this facilitates the machine-driven interpretation, integration, and validation of data. However, the resources necessary for reasoning increase exponentially with the size of ontologies under consideration. For an environment, where only limited computational resources are available, this exponential growth of processing requirements causes a scalability problem.
Real-time reasoning of semantic information with limited resources is a very important challenge in the pursuit of pervasive computing, particularly in resource-limited environments. For reasoning within some time bounds, it makes sense for the reasoner to trade correctness for computational efficiency by reducing the number of axioms initially considered. Given an ontology, the key problem we have investigated in this context is the choice of the fragment of given an ontology in order to maximize the reasoning correctness and, at the same time, the reasoning computation will finish within a given period of time.
In this dissertation, we study o...