Fuzzy ontology datatype learning using Datil

Abstract

Semantic Web technologies (especially, ontologies) are very important nowadays in intelligent applications to represent the important knowledge in an application domain. Shortly after the birth of ontologies, the inability to adequately handle knowledge affected by imprecision or vagueness, which is inherent to many real-world scenarios, was observed. Fuzzy ontologies were proposed and developed as a solution. However, the problem of building fuzzy ontologies has received little attention in the literature. In fact, there are few examples of publicly available fuzzy ontologies, suggesting that new techniques helping users to build fuzzy ontologies are needed.In this paper, we develop advanced strategies to learn fuzzy datatypes for fuzzy ontologies. Our approach uses real-world data and unsupervised clustering algorithms to compute the fuzzy datatype definitions in an automatic way. Our learning algorithm is implemented in the Datil system, a tool supporting different input formats and devices, including a mobile version. Our approach could be orthogonal to other existing approaches learning fuzzy subclass axioms.