Collaborative Spatial Decision Making Research Initiative Position Statement

Lewis D. Hopkins
May 29, 1995
University of Illinois at Urbana-Champaign
217-244-5400
907 1/2 West Nevada
Fax: 217-244-1717
Urbana, Illinois 61801
email: l-hopkins@uiuc.edu

Two areas of research would yield significant advances in collaborative spatial decision support systems:

1.Devise and test interface concepts (patterns of human computer interaction) that successfully combine representations of the substance of a problem and the processes of its exploration, particularly when multiple participants hold different concepts of the problem and play different roles in the process.

2.Consider how group processes that in practice are combinations of pure forms of group problem solving--that is, parallel processing, nominal group, collaboration, conflict resolution, voting--can be usefully, validly, and ethically supported as jointly occurring in the same interactions.

Interface Design

We have recently developed interfaces that attempt to provide both information about problem exploration processes and information about substantive solution alternatives. From a process perspective, a user should be able "do your own thing" without getting lost and without misusing any of the computer support tools through loss of validity arising from the exploration process. From the substantive perspective, a user should be able to keep track of a current alternative, previously devised good alternatives, and previously rejected alternatives, preferably at multiple levels of abstraction or completeness. Knowledge of alternatives includes not only knowledge of their structure, which is essential in considering how this structure might be changed, but also knowledge of their performance. This implies "maps" of process," maps" of alternatives, and "maps" of performance. We concur with Rasmussen et al (1994, p. 174) that "...maps support navigation in a work space more effectively than do route instructions." Our maps have taken on quite different forms, however, from theirs.

Such process-oriented interfaces have been successfully implemented in TRAINER (Johnston et al, 1994). Although we have implemented support of individual users and sessions, which permit individuals to work on problems and return to previous work, we have not yet explicitly addressed the support of collaboration in this system. The following questions arise: How should collaborators interact with a system? with each other? How should the system keep track of this collaboration? For example, can a system keep track of alternatives that have been created by two or more users independently (asynchronously?) working with a system? Or, should the system be designed to interact with the aggregate of a group process rather than individuals within it? Under what circumstances will these different strategies be effective? Should the system presume that the group rejects alternatives or that individuals reject alternatives? Should a system support multiple process views for multiple participants? How should a system move back and forth between synchronous and asynchronous activities? When and what should one participant know about the process activities or the alternatives of another participant? Can the computer help make such decisions? The general approach to interface design that we have developed should be generalizable to the collaborative case, but we will have to experiment with what collaborative patterns and tools are indeed useful.

Strong focus on process tends to submerge focus on substance, yet problem exploration by experts is based largely on consideration and manipulation of substantive representations. The addition of process support to keep track of the relationships among computer tasks and human tasks so as to ensure validity must not overwhelm focus on substance. Spatial representations of problems are used in almost all fields, whether the problem is inherently geographical or not. Thus, we should not limit the spatial representation to geographic cross sections. Spatial representations of dynamics and of substantive processes (e.g. bus route loading patterns, urban development, or animal behavior) are likely to be at least as useful in leading to recognition of new means or new alternatives. In collaborative problem exploration, it seems highly likely that a system should support both parallel processing by different members of the group and, at other times, group focus on common alternatives or issues. Can a support system help decide, for example, whether two independently created ecological models for different aspects of an ecosystem, can be validly combined? Can the data pipeline approach we have devised for individual users be generalized to multiple participants? Thus, whatever success we have achieved in representations of both substance and process, must now be tested in collaborative contexts so that appropriate data structures and system designs can be identified for groups of collaborating users.

Group processes

Group decision making is not necessarily collaborative and seldom purely collaborative. "Collaborative" Spatial Decision Support Systems must not only recognize this, but cannot be successful until we have learned more about how these group processes interact in practice. At one extreme, a team within a private consulting firm charged with developing a recommendation or set of recommendations, may come as close as any situation to pure collaboration. There is usually a project captain who has some authority to move the process forward by making decisions that are not dependent in any direct way on the opinions or attitudes of other team members. Anyone who has worked on such teams, however, knows that there are often egos, arguments, and disagreements to be worked out along the way. At the other extreme is the pure voting model in which, after whatever shared deliberation and individual investigation is conventional to a particular group, each individual casts an equally weighted vote to determine a common choice. In such cases, not only is consideration of multiple decisions (vote trading) one of the standard resolutions of Arrow's impossibility theorem, but deliberation implies interactive and common consideration of alternatives. Other models of conflict resolution beg further consideration of the many "impure" types of group processes.

One analytical tool can be used as an example to illustrate this difficulty. Benefit cost analysis is frequently used as a technical tool and as an argument for policy or legislative action. Analysts are well aware, however, that benefit cost analysis not only can be incorrectly implemented (e.g. failure to consider all resources), but also that it can be used in such a way as to make a particular argument. How might we support "collaborative" use of such evaluation tools? We have developed a design for "procedural expertise support" in which the computer support system tracks a user applying benefit cost analysis so as to catch errors in application. (The concept of "procedural expertise" is addressed more fully in Doug Johnston's submission for this conterence.) This is accomplished, for example, by the support system computing benefit cost results by multiple criteria: if net present value and benefit cost ratio do not yield the same results, then some resources have been left unaccounted for. In extending such expertise to collaborative situations, the system must also be able to comprehend differences chosen by different participants to yield different results, including the effects of such things as value measurements, discount rates, and comprehensiveness of measured effects. Such support must thus acknowledge a mixture of collaboration and conflict in the "collaborative" use of such tools.

References

Johnston, Douglas M.; Hopkins, Lewis D.; Hinrichs, Marilou; Lee, Insung; Kim, Hyong-Bok, Lu, Hsuan-Shih; and Srinivasan, Sumeeta (1994) "TRAINER: A System for Training Requirements Assessment and Integration with Environmental Resources," contract report to U. S. Army Construction Engineering Research Lab by Departments of Urban and Regional Planning and Landscape Architecture, University of Illinois at Urbana-Champaign.

Rasmussen, Jens; Pejtersen, Annelise Mark; and Goodstein, L. P. (1994) Cognitive Systems Engineering, New York: John Wiley and Sons, Inc.


Lewis D. Hopkins
Professor and Head, Department of Urban and Regional Planning
University of Illinois at Urbana-Champaign                217-244-5400
907 1/2 West Nevada                                  Fax: 217-244-1717
Urbana, Illinois 61801                             email: l-hopkins@uiuc.edu

Education: BA in Architecture (1968), Master of Regional Planning (1970), PhD in City and Regional Planning (1975), all from the University of Pennsylvania.

Employment: Assistant Professor Landscape Architecture (1972-1979), Associate Professor (1979-1984) and Professor (1984 -) Landscape Architecture, Environmental Studies, and Urban and Regional Planning, Head of Department of Urban and Regional Planning (1984 - ), all at the University of Illinois at Urbana-Champaign.

Research Interests: Human and computer problem exploration processes for incompletely defined spatial problems (spatial decision support systems, planning support systems); Land and water resources information systems (database design, dynamic geographic information systems, geographic modeling systems); Logic of urban planning (efficiency gains over time from land development planning, effects of planning on concurrency of infrastructure with development).

Research Funding: National Science Foundation, Illinois Departments of Conservation (over $1 million) and Transportation, Illinois Environmental Protection Agency, U.S. Army Construction Engineering Research Lab (over $1 million)

Editorial: Co-editor with Gill-Chin Lim, Journal of Planning Education and Research, (1987 -1991), six editorial boards.

Selected Publications

E. Downey Brill, Jr., John M. Flach, Lewis D. Hopkins, and S. Ranjithan, "MGA: A Decision Support System for Complex, Incompletely Defined Problems," IEEE Transactions on Systems, Man and Cybernetics, 20:4 (1990) pp. 745-757.

Douglas M. Johnston, Lewis D. Hopkins, Marilou Hinrichs, Insung Lee, Hyong-Bok Kim, Hsuan-Shih Lu, and Sumeeta Srinivasan, "TRAINER: A System for Training Requirements Assessment and Integration with Environmental Resources," contract report to U. S. Army Construction Engineering Research Lab by Departments of Urban and Regional Planning and Landscape Architecture, University of Illinois at Urbana-Champaign. (1994) 184 pp.

Shih-Kung Lai and Lewis D. Hopkins, "Can Decision Makers Express Multiattribute Preferences Using AHP and MUT? An Experiment," Environment and Planning B: Planning and Design, 22 (1995) pp. 21-34.

Insung Lee and Lewis D. Hopkins, "Procedural Expertise for Efficient Multiattribute Evaluation: A Procedural Support Strategy for CEA," Journal of Planning Education and Research, forthcoming.