Uniform Interpolation for Expressive Description Logics

Aus International Center for Computational Logic
Wechseln zu:Navigation, Suche

Uniform Interpolation for Expressive Description Logics

Vortrag von Patrick Koopmann
Description Logics are a family of logics that are used to model ontologies, terminological knowledge bases that are used to represent domain knowledge in a range of areas such as the semantic web, medicine, biology and many more. Uniform interpolation restricts the vocabulary used in an ontology to a specified set of terms, such that information expressible in this vocabulary is preserved. It has potentially many applications, such as in ontology reuse or ontology analysis. The theoretical properties of uniform interpolation in description logics are not encouraging for practical applications: finite representations do not always exist in classical description logics, and the size can be triple exponential with respect to the size of the input already for simpler description logics. We present an approach based on resolution that allows to compute uniform interpolants in various expressive description logics practically in a lot of cases. An evaluation on realistic ontologies shows a high success rate despite the known worst-case complexities.