Towards Practical Taxonomic Classification for Description Logics on the Semantic Web

Reference: Sanner, S. P. Towards Practical Taxonomic Classification for Description Logics on the Semantic Web. Knowledge Systems Laboratory, 2003.

Abstract: Description logics offer a well-defined semantics for many common and useful reasoning tasks that can be formalized under the notion of subsumption and are currently finding use as a representation language for the Semantic Web (e.g., DAML+OIL). For this domain, description logics can be highly useful for reasoning about relationships between entities in distributed knowledge bases by classifying them into a taxonomic subsumption hierarchy. However, there are many practical considerations for designing a sound and complete, yet efficient algorithm for performing large-scale taxonomic classification. Traditionally, structural subsumption algorithms have provided efficient techniques for performing classification but have only supported relatively inexpressive languages. Consequently, in response to the need for efficient taxonomic classification of more expressive languages, we develop an approach that extends previous structural subsumption algorithms to support sound and complete classification of a moderately expressive description logic including both conjunctive and disjunctive constructors. Furthermore, we extend this algorithm to a more expressive description logic with the claim that its sources of incompleteness are infrequent and benign in practice. Finally, we show that for the expected distribution of concept structures, both of these taxonomic classification algorithms require polynomial time in the size of the knowledge base and argue that such a result is a necessity for practical classification algorithms that will scale with the expected growth of the Semantic Web.

Full paper available as pdf.

Jump to... [KSL] [SMI] [Reports by Author] [Reports by KSL Number] [Reports by Year]
Send mail to: ksl-info@ksl.stanford.edu to send a message to the maintainer of the KSL Reports.