NCGIA Initiative 17
Collaborative Spatial Decision Making
Research Initiative Position Paper
June 1, 1995

Douglas M. Johnston
University of Illinois at Urbana-Champaign
National Center for Supercomputing Applications
4115 Beckman Institute
405 N. Mathews Avenue
Urbana, IL 61801
voice: (217) 244-5995
fax: (217) 244-4568
email: johnston@gis.uiuc.edu

Area of research:

1. Development of mechanisms to promote collaborative problem exploration through procedural support and generation of alternative solutions

2. Improvements in computational performance of spatial decision support systems to support collaboration.

Collaboration in non-computational environments focus on mechanisms of teams, conferences, and other face-to-face processes. Among the attributes of such environments are effects of immediate communication and feedback, opportunities for rapid generation of ideas, and evaluation. We assume that collaboration in spatial decision support systems should similarly be an interactive process requiring low latency and high levels of communication between human-human and human-computer processes. Barriers to these standards include diverse user requirements for decision making environments and tools, the need for multiple representations and tracking of problem explorations among participants, and delays in communication from computational demands of GIS applications.

1. Collaborative Problem Exploration.

We have worked for several years on the development of functioning experimental systems for spatial decision support. These include TRAINER: Training Requirements Assessment and Integration with Environmental Resources, a decision support system for land use allocation on army training lands; PEGASUS - an experimental urban transportation planning environment; RMS: Readiness Management System; a linked GIS - hydrologic/hydraulic model system for planning for flood events; and XCRIS: Cultural Resources Information System), a decision support system for historic and cultural resource management. While not explicitly addressing issues of collaboration, the problem domains from which the applications are drawn are focused in part on the issue of a diverse user community with possibly different stakes in the outcome of the planning process. We have sought to develop an environment for supporting collaboration by reducing dependencies on procedural knowledge of the computational environment, and on supporting the search for collectively acceptable alternatives. The aspect of collaboration we emphasize is one of problem exploration through alternative generation and multicriteria evaluation rather than consensus building or other strategies.

The search for good alternatives has been described as "an informed process of trial and error which generates alternatives and prepares them for testing" (Harris and Batty, 1994). Challenging enough, the search is made more difficult particularly when the solution space is enlarged to include multiple participants. The emphasis must be on the "informed". For example, explorations which retrace previously traveled areas must be avoided, as must trivial alternatives.

We argue that "traditional" graphical user interface strategies including devices such as menus and dialogs remain inadequate to the task of providing assistance to the user in individual tasks, let alone collaborative tasks, because they simply repackage a set of individual procedures or operations. Instead, we have experimented with the implementation of process focused interfaces in which information flows between system operations and user operations are explicitly managed. Information includes the tracing of histories of explorations by individuals or groups such that users, scenarios, attributes, and alternatives considered, are part of the information flow. (Further examination of this aspect is provided in the submission by Lew Hopkins).

In our work on the military land management system TRAINER (Johnston et al., 1994) a fundamental impediment to collaboration between the training community and land managers is the phenomena of anchoring -- specifically land resource allocation decisions based on familiarity with a particular geographic region within a particular time frame. An approach to resolving this effect is to employ it to derive initial solutions, or for comparison with alternatives derived from other means.

Part of the role of the interface is to assist in the search for sets of alternatives. For this problem, we have used a framework of a planning process similar to that suggested by Armstrong (1994) and others of strategizing, exploration, and evaluation. We are examining the problem of developing useful strategies for problem exploration. One approach is through the provision of partially substitutable procedures, that is, the user must be able to choose a strategy or model with which to pursue problem exploration. The rationale for this is that systems models may not exist for the intangible or unmodeled criteria that the user wishes to employ in the search for solutions or that the user may wish to test different strategies for generating alternatives.

We have employed Modeling to Generate Alternatives (Brill et al., 1982) to support user-driven search for alternatives by creating alternatives which are different in geographic space (and therefore varying in attribute levels) but similar in objective space (nearly equivalent alternatives). We are also testing filtering techniques to select a set of candidate alternatives from large numbers of alternatives generated across a range of attributes. We remain aware of the limitations of modeling approaches to alternative generation and have examined the use of information to trigger human strategies for alternative generation in the context of a bus route design decision support system. These and other strategies must be further examined and tested to provide effective search in a collaborative spatial decision making environment.

2. Improvements to performance features of computational GIS to support interaction between participants.

In our work we have encountered obstacles to implementation of alternative generation strategies due to computational burdens of spatial search and display of results. To achieve acceptable levels of performance, we are forced to reduce the resolution of our data, reduce the geographic extent of search, or simplify the modeling process (for example, by using model outputs rather than integrating systems models into alternative generation processes). This problem, identified by others as well (e.g. Hodgson, et al, 1995), limits the potential for collaborative environments. If we (reasonably) assume that part of a successful strategy for collaborative GIS includes rapid generation of tentative solutions (e.g. brainstorming) by varying parameter values or heuristics, and evaluations of alternatives against multiple attributes, then latency between initiation of solution strategies and results becomes a very real problem.

We are in early stages of initiaing work, in conjunction with the National Center for Supercomputing Applications, on the development of designs for scalable computational environments for modeling of spatial processes. The concept behind scalable computing is to build an infrastructure of computers ranging from workstations to large supercomputers all based on the same CPU and operating system, interconnected through high-speed networks. This design conceivably permits transparent migration of processing tasks from local workstations to GIS servers housed on supercomputers when high performance is required.

Related performance concerns involved display of spatial data. Collaboration will require multiple views of data at different temporal and spatial scales. Some models for collaboration (e.g. Shiffer, 1993) involve a conference room design with a common display. Conventional workstation displays have insufficient resolution, scale, and through-put to support the large amounts of displayed information required for collaborative planning.

Experimental settings are under development to explore display of spatial data on large scale, high resolution devices based on arrays of display drivers and display devices. While it is clear that such technologies currently lie outside the realm of common application, this work will allow experimentation with different models for collaboration.

REFERENCES

Armstrong, M. 1994. Requirements for the Development of GIS-Based Group Decision Support Systems. Journal of the American Society for Information Science, 45(9):669-677.

Brill, E.D., S.Y. Chang and L.D. Hopkins. 1982. Modeling to Generate Alternatives: The HSJ Approach and an Illustration Using a Problem in Land Use Planning. Management Science, 25(2): 221-235.

Harris, B. and M. Batty. 1993. Locational Models, Geographic Information Systems and Planning Support Systems. Journal of Planning Education and Research 12:184-198.

Hodgson, M.E., Y. Cheng, P.Coleman,and R.Durfee. 1995. Computational GIS Burdens. GeoInfo Systems, April, 1995: 29-37.

Johnston, D.M., L.D. Hopkins, M. Hinrichs, I. Lee, H-B Kim, H-C Lu, and S. Srinivasan. TRAINER: A System for Training Requirements Assessment and Integration with Environmental Resources, Department of Landscape Architecture, Department of Urban and Regional Planning, University of Illinois Urbana, Illinois (1994) 184 pp.

Shiffer, M.J. 1992. Towards a Collaborative Planning System. Environment and Planning B: Planning and Design. 10:709-722.


DOUGLAS M. JOHNSTON

Associate Professor, Department of Landscape Architecture
Director, Geographic Information Systems Laboratory
Senior Research Scientist, National Center for Supercomputing Applications

214 Mumford Hall        			(217) 244-5995
1301 W. Gregory Drive   			(217) 244-4568 (fax)
University of Illinois at Urbana-Champaign      dmj@ncsa.uiuc.edu
Urbana, Illinois  61801

Education: B.S., B.L.A.(1979,1980), SUNY College of Environmental Science and Forestry; M.L.A. (1982) Harvad; Ph.D. (1986) University of Washington (Civil Engineering)

Selected Publications

Johnston, D.M., and R.N. Palmer. "Application of Fuzzy Decision Making: An Evaluation." Civil Engineering Systems. 5(2), (1988). pp. 87-92.

Stuebe, M., and D.M. Johnston. "Hydrologic Modeling Using GIS." Water Resources Bulletin. 24(4), (1990) pp. 1-10.

Hopkins, L.D. and D.M. Johnston. "Locating Spatially Complex Activities with Symbolic Reasoning." Proceedings of the 4th International Symposium on Spatial Data Handling, (1990) pp.762-771

Frederickson, K. and D.M. Johnston. "A GIS-Hydrologic Model Interface for Flood Prediction and Assessment." (1993)

Johnston, D.M., L.D. Hopkins, M. Hinrichs, I. Lee, H-B Kim, H-C Lu, and S. Srinivasan. TRAINER: A System for Training Requirements Assessment and Integration with Environmental Resources, Department of Landscape Architecture, Department of Urban and Regional Planning, University of Illinois Urbana, Illinois (1994) 184 pp.

Selected Grants Received

1986-1989 "Illinois Streams InformationSystem (ISIS)." With L.D. Hopkins (Principal Investigator). Illinois Department of Conservation. $100,000 - 200,000 /yr.

1988 "Prototypical Land Condition-Trend Analysis Database System." With David Kovacic. U.S. Army Construction Engineering Research Laboratory. $117,000.

1990-1995 "Development of an Army-Unique Approach to Land Use Allocation Strategies for Training Lands." with Lewis D. Hopkins. U.S. Army Construction Engineering Research Laboratory. $70,000/yr.

1991-1992 "Prototype Geographic Regulatory Information Program Development". Principal Investigator. Illinois Environmental Protection Agency. 1991: $48,200, 1992: $46,361.

1992 Develop a Decision Support System for Cultural Resource Management. U.S. Army Construction Engineering Research Lab. Principal Investigator. $59,577

1994 Development of a GIS Interface to the Great Lakes Water Control Data System. U.S. Army Corps of Engineers, Detroit District. In conjunction with the U.S. Army Construction Engineering Research Lab. $50,000

Areas of Research

Environmental Planning. Use of spatial and temporal models for assessment and allocation for environmental planning problems. Development of spatial decision support systems, for example for flood management and military land management. Use of geographic information systems to facilitate public education and involvement in social and environmental problems.