Finding informative commonalities in concept collections

The problem of finding commonalities characterizes several Knowledge Management scenarios involving collection of resources. The automatic extraction of shared features in a collection of resource descriptions formalized in accordance with a logic language has been in fact widely investigated in the past. In particular, with reference to Description Logics concept descriptions, Least Common Subsumers have been specifically introduced. Nevertheless, such studies focused on identifying features shared by the whole collection. The paper proposes instead novel reasoning services in Description Logics, aimed at identifying commonalities in a significant portion of the collection, rather than in the collection as a whole. In particular, common subsumers adding informative content to the one provided by the Least Common Subsumer are here investigated. The new services are useful in all scenarios where features are not required to be fully shared, like the one motivating our research: Core Competence extraction in knowledge intensive companies