The Semantic Web is a collection of many different data pages. It is still unclear how to answer a query posed on the Semantic Web using formal methods. Currently, the main approach to the above problem is addressed by description logics. However, description logics treat the Semantic Web as a single assertional knowledge base, based on a single ontology. In general, many web pages are irrelevant for a given query. Consequently, data retrieval services in description logic systems can be inefficiency if straightforward retrieval algorithms are chosen.
In this work, we propose a space reduction methodology to address this issue. In particular, we develop techniques aimed at the reduction of the search space a description logic reasoning algorithm needs to take into account for answering a query. To reduce the search space, we need to be able to identify 1) the dependency between pairs of Semantic Web data sources in order to maintain soundness of reasoning, and 2) the data sources which are fully irrelevant. Thus we need a way to determine whether a data source is relevant with respect to a query. Consequently, each data source must be associated with its source description. We propose a specification of source description and show how it can be used to reduce the reasoning search space.
It has been argued that reasoning on the Semantic Web will benefit from the addition of rule systems to description logics. Accordingly, we specify how to combine rule systems, in particular defeasible logic, to description logics, and compare the expressivity of the rule system and the description logic system. We choose defeasible logic since, to the best of our knowledge, it is the only non- monotonic reasoner that can operate in PTime. A nonmonotonic rule system gives us the ability to handle incomplete information in an easier way than description logic systems. We also extend the proposed space reduction method to the logic resulting from the combination of defeasible and description logic, defeasible description logic (DDL).
The results from this PhD research allow one to ﬁnd the answer of a complex query from a description logic-based, single ontology, Semantic Web system. Furthermore, the efficiency of description logic-based query answering also beneﬁt from this research, and so the efficiency of rule-over description logic-based query answering.