A Declarative Modeling Language for Concept Learning in Description Logics

Learning in Description Logics (DLs) has been paid increasing attention over the last decade. Several and diverse approaches have been proposed which however share the common feature of extending and adapting previous work in Concept Learning to the novel representation framework of DLs. In this paper we present a declarative modeling language for Concept Learning in DLs which relies on recent results in the fields of Knowledge Representation and Machine Learning. Based on second-order DLs, it allows for modeling Concept Learning problems as constructive DL reasoning tasks where the construction of the solution to the problem may be subject to optimality criteria