Martin Raubal, University of California, Santa Barbara

Abstract—Geographic information science provides the foundation for the development of geospatial tools and services that support people in their spatio-temporal decision-making. In order to offer useful and useable solutions, principles of human spatial cognition regarding the representation and processing of spatial and temporal aspects of phenomena must be considered in the design of these tools. Such cognitively engineered geospatial services aim for cognitive adequacy and therefore facilitation of user interaction. This article argues for the necessity of cognitive engineering methods in the field of geographic information science by explicating their theoretical foundation and demonstrating practical geospatial applications. It further provides a framework for classifying cognitive user parameters, which can be employed for the personalization of geospatial services.

Conclusions and Future Research Directions

Geographic information science investigates all aspects of geographic information. It therefore provides the scientific foundation for geospatial tools and services. In order to be useful and usable for people, these tools must be designed and engineered by incorporating principles of human cognition with respect to spatial and temporal aspects of phenomena. Only then will they facilitate human-computer interaction and provide high-quality support for their users' spatio-temporal decision-making. This article argued for the importance of integrating cognitive engineering methods within the field of GIScience. Cognitive engineering is a type of applied cognitive science and has its roots in user-interface design. Geospatial services are unique in the way they incorporate spatial and temporal data, therefore requiring the extended version of spatial cognitive engineering. Its major goal is personalization of geospatial services because users of geographic information differ in their cognitive styles, abilities, and preferences. Based on research in the cognitive sciences and human computer interaction – more specifically in psychology, spatial cognition, and user interface design – we provided a framework for classifying cognitive user parameters into generic, group, and individual. Such parameters concern people's concepts, abilities, and strategies, and must be utilized for the personalization of geospatial tools and services. Two application areas from mobile decision-making were chosen as examples

for such cognitively engineered geospatial services. They demonstrated that spatial cognitive engineering supports all three levels of personalization. First, geospatial services must account for universal concepts and strategies in order to come closer to their users' thinking in general and to be based on solid cognitive ground. Second, by taking specific cognitive characteristics of groups into consideration, the systems' solutions can be improved for each individual group. Finally, personalization of geospatial services leads to the highest level of usefulness and usability for the individual user but it is hardest to achieve because of the sometimes unlimited complexity regarding data acquisition, modeling, and representation.

The area of spatial cognitive engineering leads to a series of research directions for future investigation:

• Formal Conceptual Representations. People's conceptual representations need to be formalized in order to integrate them into computer systems. The closer the system's view comes to the user's view in terms of conceptualizations and reasoning processes, the higher the likelihood of facilitating human-computer interaction and delivering cognitively adequate answers to the users' spatio-temporal problems. Various formal approaches to cognitive modeling, and the representation and processing of geographic knowledge exist (Barkowsky 2002), resulting from different views on the nature of conceptual representations in the human cognitive system. This leads to the question whether there is a hybrid spatial cognitive model that covers 'the whole ground' and if so, what are its components? Multidisciplinary research in the cognitive sciences, the geosciences, and computer science is required to find answers to this question. Human participants tests may help assess the validity and potential limitations of conceptual representations. Cognitively based spatio-temporal ontologies, such as the Descriptive Ontology for Linguistic and Cognitive Engineering (DOLCE) (Gangemi et al. 2002), can be used as a foundational representational upper-level framework for specifying such conceptualizations.

• Spatio-Temporal Decision-Making. Usefulness and usability of cognitively engineered tools depend strongly on how well their underlying theories represent elements of people's spatial and temporal decision-making, including aspects of uncertainty and vagueness in the real world. It is necessary to investigate the various effects of geographic and temporal scales on people's information requirements. Such effects will help to establish different levels of granularity for the represented information. Further research is necessary to explore how time constraints for different types of geographic information and desired activities are constructed and how this process is influenced by cultural differences.

• Representing Context. Geospatial information must be understood by different users and in different contexts. The semantics of expressions for objects, actions, and relationships can change significantly depending on the user's background, perspectives, and situation. Investigations on people's different foci with respect to their activities in space and time will help to establish sets of parameters for focalization, i.e., the adaptation to different decision situations (Raubal and Panov 2009). An example is the modeling of different foci onto urban space during wayfinding. Furthermore, context must be quantified in order to be utilized in geospatial services. One possible way of quantification is to assign weights to measures, depending on the cognitive representation. Human participants tests will help to determine these weights for different context parameters.

•
Applications and Evaluation. A large number of different models and prototypes for geospatial services exists, in addition to the applications discussed in this article. In order to test their usefulness and usability in a comprehensive way, they have to be fully implemented and further extended. This also concerns architectural and user interface issues. Not only is it important whether such services are technically feasible but with the help of human participants tests one also needs to find out how well they work for their users and whether or not people want to use them in their everyday lives.

Reference：

Martin R. Cognitive Engineering for Geographic Information Science [J]. Geography Compass, 2009, 3(3):1087-1104.