A Pattern-based Framework for Representation of Uncertainty in Ontologies
Miroslav Vacura, Vojtěch Svátek, Nick Simou · 2007
Abstract. We present a novel approach to representing uncertain information in ontologies based on design patterns. We provide a brief description of our approach, present its use in case of fuzzy information and probabilistic information, and describe the possibility to model multiple types of uncertainty in a single ontology. We also shortly present an appropriate fuzzy reasoning tool and define a complex ontology architecture for well-founded handling of uncertain information. Motivation for our research is the CARETAKER project 4 which comprises advanced approaches to recognition of multimedia data, which led us to problems of representing uncertain information. Although fuzziness isn’t, exactly said, type of uncertainty, we will in this example consider representing fuzzy information in the form of facts, i.e. A-Box from description logic (DL) point of view. The key principle of our approach to representing fuzzy information is the separation of crisp ontology from fuzzy information ontology. We allow the fuzzy ontology to be OWL Full and only suppose that the base ontology is OWL DL compliant. Regular OWL DL crisp reasoning tools can be applied to the base ontology, fuzzy reasoning tools (i.e. FiRE 5) to fuzzy ontology. Instantiation axioms in Fuzzy OWL [1] are assertions of form 〈a: C ⊲ ⊳ n 〉 – facts saying that individual a belongs to class C, n is level of certainty (0, 1) and ⊲ ⊳ is one of {≤, }. We introduce a few constructs that enable us to model such axioms with uncertainty by ontology patterns. For each crisp axiom of base ontology we create a new individual belonging to class fuzzy-instantiation, which will have several properties attaching it to that crisp axiom in base ontology and implementing uncertainty. Properties fi-instance and fi-class characterize the membership of an individual person-1 to class problem-person. Property f-type defines the type of uncertainty relation (⊲⊳) and datatype property f-value defines the level of uncertainty n (Fig. 1, individuals are grayed and classes are bright).