A GeoAgent‐based framework for knowledge‐oriented representation: Embracing social rules in GIS

Abstract

While current Geographic Information Systems (GISs) can represent observational spatial data well, they have limited capabilities in representing some non‐observational social elements and goal‐driven behaviours that can be important factors in a wide range of geographic issues. Such social components may include laws, regulations, polices, plans, culture, and customs, as well as their relations and interactions with the geographic environment at different scales. Getting beyond traditional data‐centred approaches, this research presents a knowledge‐oriented strategy in order to address these issues within a GIS context. We incorporate two major conceptual elements. First, extending from conventional agent notions and their geographic applications, geographic agents (GeoAgents) are considered as a basic representation component to specifically address social rules and goal‐driven behaviours that impact the Earth and environmental systems. Second, in order to incorporate GeoAgents with current space–time representation, a new conceptual representation framework, called ‘fields, objects, time, GeoAgents, and relations’ (FOTAR), is introduced to address the cross‐scale processes of both social and natural interactions. A Java‐based prototype, GeoAgent‐based Knowledge System (GeoAgentKS), is described to implement this framework by integrating agent technologies with multiple data and knowledge representation techniques, such as expert systems, concept maps, mathematical models, and geospatial databases. The application of this prototype in a case study is also presented, investigating scale‐dependent human–environment interactions under different emergency situations for community water systems in Central Pennsylvania, USA. In this case study, a systematic set of methodologies of knowledge acquisition, representation, and confirmation for constructing GeoAgents' knowledge bases by using expert systems were explored to formalize high‐level knowledge and social behaviours in the FOTAR‐based representations. The results show that the proposed conceptual representation framework is achievable at both implementation and application levels, and the prototype tool is demonstrated to be valuable in facilitating knowledge sharing, policymaking, municipal management, and decision‐making, especially for real‐world emergency management.

Keywords:

• Representation
• GeoAgents
• Social rules
• Human–environment interactions
• Expert systems
• Emergency management

Acknowledgements

We wish to acknowledge the joint support from the Human–Environment Regional Observatory (HERO) project by the National Science Foundation, the National Oceanic and Atmospheric Administration (NSF Grant SBE‐9978052, Brent Yarnal, Principal Investigator) and the GCCM (GeoCollaborative Crisis Management) project (NSF grant EIA‐0306845, Alan MacEachren Principal Investigator). We are grateful for the insightful comments from Brent Yarnal, Alan MacEachren, Mark Gahegan, and the anonymous reviewers. Thanks also go to the help from David O'Sullivan, Junyan Luo, Xinhua Han, and other members in the GeoVISTA Center and the HERO research team. Human subject protection for social science research is under IRB No. 15986 at the Pennsylvania State University.