The Modular Structure of an Ontology: Atomic Decomposition

Modularity in ontologies Modern ontologies can get quite large as well as complex, which poses challenges to tools and users when it comes to processing, editing, analyzing them, or reusing their parts. This suggests that exploiting modularity of ontologies might be fruitful, and research into this topic has been an active area for ontology engineering. Much recent effort has gone into developing logically sensible modules, that is, parts of an ontology which offer strong logical guarantees for intuitive modular properties. One such guarantee is called coverage. It means that a module captures all the ontology’s knowledge about a given set of terms (signature). A module in this sense is a subset of an ontology’s axioms that provides coverage for a signature, and each possible signature determines such a module. The minimal modules to provide coverage for a signature are those based on Conservative Extensions (CEs) [2], that are however not feasible to be computed for many expressive languages. Modules based on syntactic locality [5] also provide coverage because they are efficiently computable approximations of CEs; however, such modules are not in general minimal. The extraction of such a module given a set of terms (signature) is well understood and starting to be deployed in standard ontology development environments, such as Protégé 4,1 and online.2 Locality-based modules have already been effectively used for ontology reuse [14] and as a subservice for incremental reasoning [3]. However, we think that by investigating the family FO of all locality-based modules we can obtain information about topicality, connectedness, structure, superfluous parts of an ontology, or agreement between actual and intended modeling.