Reasoning with Large A-Boxes in Fuzzy Description Logics using DL reasoners: An experimental evaluation

While knowledge representation languages developed in the context of the Semantic Web build mainly on crisp knowledge, fuzzy reasoning seems to be needed for many applications, e.g. for retrieval of images or other resources. However, techniques for fuzzy reasoning need to scale to large A-Boxes for practical applications. As a first step towards clarifying whether fuzzy reasoning techniques can indeed scale, we examine a specific point in space. Earlier research has shown that fuzzy description logics can be transformed to crisp description logics in a satisfiability-preserving fashion. This thus opens the possibility of using standard description logic reasoners for reasoning with fuzzy OWL ontologies. As the transformation produces a quadratic blow-up of the T-Box, a crucial question is if such an approach is feasible from a performance point of view. We provide an empirical evaluation on four different ontologies of varying complexity and with generated A-Boxes of up to a million individuals. To our knowledge, we thus provide the first systematic and empirical evaluation of a fuzzy reasoning approach based on a reduction to standard description logics.