Semantic Interoperability and Cultural Specificity

Creating a database of historical Chinese administrative units for the CHGIS project required the definition of specialized feature classifications. Although limited to the domain of Chinese language materials, these classifications revealed variability over time, region, and ethnic or cultural dimensions. To explore the variable usage of feature classification terminology across linguistic and regional boundaries, the official cartographic feature classification systems for China, Taiwan, and Japan were integrated with the CHGIS historical types by mapping the types from all of the separate lists to those found in a designated control vocabulary. Here the process is extended to include traditional Mongolian and Uighur geographic feature types. The culturally specific terms found in the Mongol and Uighur entries demonstrate the general truth, which is that the majority of existing geographic classification systems remain highly unstructured, ad-hoc agglomerations of terms, thrown together for the expedient purposes of each individual project. Not until sufficient accumulation of geographic terms across many languages and disciplines is done, will we be able to propose the deep structure of a truly generic ontology for geographic features. 1.1 Overview of Ontologies The specific approaches to semantic interoperability among geographic feature classification systems addressed in this paper will make more sense if we begin by stepping back a few paces to take an overview of the general terminology and concepts being used in the field today. The bulk of this overview is drawn from the work of a leading center of expertise, Ontogeo, where many years of research have already been completed. [Ontogeo] The gist of Ontogeo’s research has been to discover the steps in the process of integrating various classification systems, and to come up with a formal language and methodology for doing so. Ontologies have been recently defined as “theories that use a specific vocabulary to describe entities, classes, properties, and functions related to a certain view of the world. They can be a simple taxonomy, a lexicon or thesaurus, or even a fully axiomatized theory.” [Fonseca, 2002] What differentiates an ontology from a thesaurus is that the ontology is not only devised as a means of relating existing terms to one another, but seeks to define specific concepts within a domain of knowledge, and to define the relationships between those concepts. [Beck, 2002] 1.2 Informal to Formal Ontologies The ontologies which we commonly come across have been formulated for various purposes, and don’t conform to any particular standard. Therefore we differentiate them according to several types ranging from the informal to the formal. The simpler taxonomical types are called informal ontologies, and generally exist as a list of terms, arranged in hierarchical relationships. If natural language definitions are provided for each term in the hierarchy, they are called terminological ontologies. When the relationships between each term have been explicitly stated in formal logic or as axioms that can be processed programmatically, we are dealing with formal ontologies. Unfortunately or fortunately for us, most of the ontologies which are currently in use for describing geographic features are very informal, ad-hoc collections of terms which were devised for a specific purpose. My position, explained in the following sections, is that we can take better advantage of informal ontologies at first, as we build up a network of related concepts, and that this work is prerequisite to establishing finer distinctions between classes and individual concepts.