Experiments to Examine the Situated Nature of Geoscientific Concepts

ABSTRACT

Ontologies are being developed in many geoscientific domains. They are typically populated with two types of concepts: upper-level concepts that apply across many or all domains, and domain concepts that apply only within a specific domain. Previous work has refined this distinction by identifying a particular type of domain concept, called a situated concept, which is dependent on specific processes (natural, social, scientific, or possibly machine) for its meaning and is instantiated amongst entities within a specific geographical and historical context. In this paper we present new support for situated geoscientific concepts, building on our previous research that argues for the importance of situations in the development and use of concepts related to geoscientific field mapping. The new results are obtained by using statistical techniques to further analyze three geologists' field data over time, to better test the hypothesis that the concepts developed by the geologists to classify objects on the map are in fact situated. The field data are compared to each other, and to the concepts developed by the team. Differences found between and within individuals' data for three map concepts provide strong support for the idea that the concepts are variably influenced by data, theory, and natural and human situations. From this increased corroboration of situated concepts we suggest two implications for domain ontologies: (1) a delineation between situated domain concepts and non-situated domain concepts; and (2) recognition that representation of reliable meaning involves the capture of historical and geographical context for situated concepts.