|Title||Resolving semantic conflicts through ontological layering|
We examine the problem of semantic interoperability in modern software systems, which exhibit pervasiveness, a range of heterogeneities and in particular, semantic heterogeneity of data models which are built upon ubiquitous data repositories. We investigate whether we can build ontologies upon heterogeneous data repositories in order to resolve semantic conflicts in them, and achieve their semantic interoperability. We propose a layered software architecture, which accommodates in its core, ontological layering, resulting in a Generic ontology for Context aware, Interoperable and Data sharing (Go-CID) software applications. The software architecture supports retrievals from various data repositories and resolves semantic conflicts which arise from heterogeneities inherent in them. It allows extendibility of heterogeneous data repositories through ontological layering, whilst preserving the autonomy of their individual elements.
Our specific ontological layering for interoperable data repositories is based on clearly defined reasoning mechanisms in order to perform ontology mappings. The reasoning mechanisms depend on the user‟s involvments in retrievals of and types of semantic conflicts, which we have to resolve after identifying semantically related data. Ontologies are described in terms of ontological concepts and their semantic roles that make the types of semantic conflicts explicit. We contextualise semantically related data through our own categorisation of semantic conflicts and their degrees of similarities.
Our software architecture has been tested through a case study of retrievals of semantically related data across repositories in pervasive healthcare and deployed with Semantic Web technology. The extensions to the research results include the applicability of our ontological layering and reasoning mechanisms in various problem domains and in environments where we need to (i) establish if and when we have overlapping “semantics”, and (ii) infer/assert a correct set of “semantics” which can support any decision making in such domains.