SOME CHALLENGES AND APPROACHES TO REPRESENTING GEOSCIENTIFIC KNOWLEDGE IN CYBERINFRASTRUCTURE

Networks of databases and software tools are being constructed in the geosciences as part of an emerging cyberinfrastructure. These networks are responding to (1) the need to access high performance computing resources and data stores remotely, (2) increases in the size, complexity and heterogeneity of data, and (3) the need to share computational resources such as data, tools, methods, hypotheses and results more effectively. Underpinning these needs is a widespread vision that novel use of cyberinfrastructure will lead to new research methods and results. Expected benefits include improved interoperability of information and software, as well as augmented capacity to formulate and test hypotheses against the digital rock record. Because realization of these benefits depends on adequate representation and manipulation of geoscientific knowledge, e.g. to resolve diverse database and mental schema, we explore the nature of such knowledge and the suitability of representing it using computational ontologies. Identified are several challenges to representation arising from the dynamic, partial, plural and situational character of geoscientific knowledge. Also presented is an approach that builds on current ontology frameworks to address some of the challenges. A wider implication arising from the challenges and approach is the notion that cyberinfrastructure should facilitate the management of explicitly represented geoscientific knowledge, including multiple ontologies and models, and should also attempt to manage aspects of knowledge that at present remain implicit (such as situations and context) as well as the theoretical, methodological, and conceptual assumptions surrounding them. This should not only facilitate scientific activity within cyberinfrastructure, but also help steer the infrastructure toward its presumed destiny of becoming a new operational paradigm for the geosciences.