Ontology‐Based Metadata

Abstract

It is even more remarkable to some of us that up to this point, there has been no thoroughgoing discussion by GIS practitioners and theorists about the epistemology of their subject, the ontology of their objects, and the political commitments embedded in their practices.(Pickles 1997, p. 364) AbstractMetadata are an extant mechanism for conveying ontological information about semantic data. Metadata have the advantage of being institutionally and structurally ensconced in GIS. At present, however, they lack fields to express information beyond the technical and geometric domain. This paper describes a framework for the creation of extended metadata for non‐spatial attributes, with the goal of incorporating ontological context. We use an informatics interpretation of ontology which refers to the total universe of discourse associated with a given attribute (database field). In other words, an ontology is the possible range of meaning offered by an encoded field. Current ontology research in GIScience has focused on data structuring and modelling. Implementation of these schemes demands restructuring of existing relational database models. An alternative is to extend current metadata schemes (e.g. ISO 19115) to include context‐based and tacit information about semantic attributes. Such ontology‐based extended metadata permits data selection and interoperability decisions that are ultimately more defensible. We have developed eight preliminary fields to add to existing metadata frameworks that will enable ontological context to travel with the data. This paper illustrates a preliminary implementation based on the integration of non‐commensurate cadastral data. The results illustrate the value of ontology‐based metadata in highlighting descriptive and substantive differences between similar classification systems. Recognition of semantic heterogeneity is the basis for creating defensible data linkages between multiple datasets. The development of ontology‐based metadata is profoundly different from the current trend to incorporate ontological context at the model level in GIScience. It is pragmatic, however, in that it presents a vehicle for incorporating use‐context with data in a manner that is accessible; requires little re‐engineering; and is intuitively understood by GIS users. Moreover, ontology‐based metadata are a mechanism for addressing Pickles’ (1997) concern about the ontology of spatial objects in GIS.