Panel chair and members
The term Computational Implementations is understood here in a broad sense, ranging from formalizations of geographic concepts to considerations about the development of effective software systems. Effectiveness includes the improved communication between users and systems, the availability of innovative GIS technologies tailored to the users’ tasks, and the integration and interoperation of system components to best serve the analysis of novice users. Most current methods in geographic information science were designed from the perspectives of the computer scientist and the cartographer, aiming at efficiency in capture, storage, and processing of cartographic features. The state of the art in formalizations of geographic knowledge, as reflected in most current GISs, requires that certain constraints be fulfilled before a user is allowed to perform any analysis. They include that positions must be recorded in absolute terms in a Cartesian coordinate space, geographic objects must be described by precisely defined boundaries, and all geographic data sets must be complete. Our goal is to overcome some of these and other limitations, and to find formal representations that come closer to human practice, capture more complex geographic concepts, and better match cognitive processes.
2. Non-Varenius events related to Computational Implementations of Geographic Concepts, since October 1995
Fifth International Symposium on Large Spatial Databases
This bi-annual conference has become the most significant international meeting with respect to computational implementations. The fifth meeting of this series was held in July 1997 in Berlin with approximately 100 international participants. From among over 50 submissions of full papers, 19 were selected for the program. Three of the five panelists served on the program committee. The major topics included Spatial Similarities, Spatial Constraint Databases, Spatial Query Processing, Spatial Data Models, and Spatial Access Methods. SSD '99 will be held in Hong Kong.
Since this year, the area of computational implementations has its own journal with the publication of GeoInformatica (Kluwer). The Panel Chair is a co-editor and two other panelists and the project Director are members of the editorial board.
Within the realm of Computational Implementations, the Panel endorsed three initiatives, which will hold their specialist meetings between December 1997 and February 1999: Interoperating Geographic Information Systems, The Ontology of Fields, and Discovering Geographic Knowledge in Data-Rich Environments. Three of the five Panel members attended the Panel's meeting in Santa Barbara, May 2–3, 1997. Over the year, several ad hoc meetings with panel members were arranged at various conferences. At this point in time, the initiative on Interoperating Geographic Information Systems is close to its Specialist Meeting; the initiative on the Ontology of Fields is close to finalizing the core planning group and the call for participation; and we are actively recruiting the leaders for the initiative on Discovering Geographic Knowledge in Data-Rich Environments.
INTEROPERATING GEOGRAPHIC INFORMATION SYSTEMS
LEADERS:
Location: Santa Barbara, CA
Interoperability offers one possible way of making GIS more useful and accessible to scientific research by making interactions between users and GISs easier and obviating the need for complex techniques to overcome incompatibilities between software systems and data sets. While much attention has been devoted to the question of GIS interoperability in recent years, less attention has been paid to the theoretical and methodologically sound basis for a new generation of component-based interoperating processes. Research topics under this initiative will include:
THE ONTOLOGY OF FIELDS
LEADERS:
Location: Coast of Maine
While much attention has been devoted to understanding the ways people conceptualize geographic phenomena as discrete objects, the alternative conceptualization as spatially continuous fields has received much less attention from a cognitive perspective. Fields are widely used as a scientific concept, particularly in mathematical physics, and many geographically distributed variables (e.g., elevation and temperature) are conceptualized as single-valued functions of location. In everyday discourse, however, we have a comparative paucity of terms to describe continuous variation, and appear to favor descriptions that replace fields with discrete objects (peaks, valleys, fronts). Computer representations similarly favor discrete objects, and force spatially continuous fields to be discretized. Is human cognition indeed less accommodating to field concepts? What are the functional interrelationships between object and field types of representations in human cognition? How can field representations be accommodated within contemporary paradigms of computing? How can the cognitive interrelationships be operationalized? What methods can be devised to measure the effects of discretization? What options exist for representing uncertainty in fields, and are they meaningful from a cognitive perspective? To address these questions, input from the disciplinary communities of GIS, Philosophy, Mathematics, Cognitive and Developmental Psychology, Image Interpretation and Image Understanding, Spatial Statistics and example geographic domains (e.g., Climatology, Soils Science, Urban Geography) must participate and interact. Some specific topics include:
LEADERS: tba (from image processing, spatial statistics, spatial database systems, or digital libraries communities)
Core planning group:
Location: Santa Barbara, CA
Digital geographic datasets are growing exponentially and under such activities as the development of the National Spatial Data Infrastructure, the launching of new satellite systems with higher resolutions, and the day-to-day collection of digital imagery, video, and sound. Society has changed from being data-poor to data-rich, while our techniques for deriving knowledge from the data in an analytical context have remained inferential in nature. The problem has now become not finding the data, but filtering through large volumes of data to finding meaningful geographic knowledge. At the same time, the types of datasets available are changing from the traditional vector and raster sets, to include such data types as video and audio, and the location of where these data were collected. We must overcome these limitations and develop new approaches and methods that focus upon separating the relevant from the irrelevant, the meaningful from the background noise. The goal of this initiative is to find new automated methods for filtering large amounts of raw geographic data into more user-consumable forms of knowledge. This includes:
Over the last two years, the area of computational implementations has gotten significant attention in GIS research and development, both within the academic research community as well as with industry. Probably the most significant changes are currently influenced by the formulation and implementation of the Interoperability Specifications of the Open GIS Consortium (http://www.opengis.org). The Varenius initiative on "Interoperating Geographic Information Systems" with the International Conference and the Specialist Meeting (both to be held in December 1997) will provide a complementary focus for activities that occur in relation to OGC's efforts.
In a similar vein, the development of the ISO Standard on Geographic Information/Geomatics (TC 211) has been moving forward, with many of its over 20 parts close to committee draft. Two members of this Varenius panel (Egenhofer and Herring) have been involved in the ISO work.
The GIS industry has seen a boost of support with new supporting products at the backend of GISs (mostly for data management) and many desktop products tailored to particular data processing tasks. The supporting products include the Spatial Data Engine from ESRI, the Spatial Data Option from Oracle, and Spatial Data Blades from Informix. This trend is significant as it gets additional new players such as Oracle and Informix into the broader field of GIS. Many of the desktop products come from small software companies.
We continue to observe a trend in the formalization of spatial concepts that are cognitively plausible. There have been several workshops with the goal to bring together formalists and cognitive scientists to develop research agendas (NCGIA’s I-21, COSIT ‘97, AAAI workshops ‘96 and ‘97). These activities reflect the interests in computational implementations as they often provide the stimulating momentum for new approaches to research. The symbiosis at this time includes formal-methods people in GIS, cognitively-oriented geographers, psychologists, and researchers from Artificial Intelligence.