A Practical Approach for Computing Generalization Inferences in EL

We present methods that compute generalizations of concepts or individuals described in ontologies written in the Description Logic EL. These generalizations are the basis of methods for ontology design and are the core of concept similarity measures. The reasoning service least common subsumer (lcs) generalizes a set of concepts. Similarly, the most specific concept (msc) generalizes an individual into a concept description. For EL with general EL-TBoxes, the lcs and the msc may not exist. However, it is possible to find a concept description that is the lcs (msc) up to a certain role-depth. In this paper we present a practical approach for computing the lcs and msc with a bounded depth, based on the polynomial-time completion algorithm for EL and describe its implementation.