ByRené F. Reitsma
INTRODUCTION:
ENVIRONMENTAL DECISION MAKING
Due to steady increases in the `pressure' on environmental resources on the one hand and a growing awareness of the potential values of those resources for a wide range of interests on the other, environmental conflicts and decision making have become an integral part of modern societies.
The very nexus of these environmental conflicts arises from a combination of two factors. First, discrepancies between supply and demand with respect to any kind of resource, imply conflictive situations, since scarcity of resources implies competition for those resources.
Note that the term `resources' should be interpreted here very loosely. Effectively, an environmental resource can be every aspect of the environment that can be put to use for a specific purpose. In a drought situation, for instance, the `resource' characteristics of water become very obvious. Lots of people and ecosystems need the water and if there is not enough available to supply that demand, a conflict arises. In cases of, for example, nuclear waste management, land-use planning, or flood control, the supply/demand dimension of a problem might be less obvious, although the same principles apply. If, under normal supply situations the demand increases, the potential for conflict and managerial crises increases also. Likewise, if in a situation of equilibrium the supply decreases but the demand does not, conflicts are imminent.
A second source of conflict has to do with the multiobjective nature of the demand for the resource. Typically, various interests are associated with the environmental resource under dispute. These interests are furthermore often represented by different groups of `stake-holders'. Water resources management in the Western parts of the U.S. provides many examples of this. The water from the Colorado River, for instance, is put to many different kinds of uses, both consumptive and non-consumptive: urban and industrial use, recreational facilities such as rafting and sports fishing, or the generation of electricity. Moreover, the river water also serves fish hatcheries as well as many environmental purposes.
The nature of these kinds of situations creates a significant potential for conflicts among various groups of water users. Depending on the nature of the actual problem, multiobjective environmental problems tend to be zero-sum problems; having winners implies having losers. In water resources management problems this zero-sum conflict situation applies to both consumptive and non-consumptive use of water. Consumptive use, of course, implies that water once used is gone from the system (we do not concern here the natural processes of return flow, groundwater aquifer storage, etc.). The Burgesses mention that the zero-sum situation becomes very apparent in cases of fully appropriated water allocation systems. Although less obvious, similar relationships hold for non-consumptive uses also. For example, rafting on the Colorado River requires certain stream flows, such that rafting is possible and that, during a multi-day trip, rafters can camp on the river embankments at night. However, electricity generation through hydro-electric power plants can influence stream flow regimes dramatically. Therefore, unless timing of non-consumptive uses is perfect, these uses may very well impair each other. Another example of conflicts in interests associated with the `transport' dimension of water is the change in streamflow regimes through the release of water for the generation of electricity. These releases significantly influence processes such as sedimentation, thus endangering the natural build-up or even washing away beaches and embankments.
Sometimes trade-off relations between objectives exist so that `internal' compensations can be worked out. More often, however, damage to one objective cannot be easily compensated by gains in another objective and compensation, if at all, can only be carried out through the use of a `universal' compensator, i.e., money. However, in most cases this is only a hypothetical possibility. Much more frequently disputes are settled by forcing `consensus' among disputees using techniques such as voting, litigation, or, less civilized means of pressure. A `worst case' scenario is one in which decision making is postponed beyond the point at which the damages to interests could, in principle, be kept within reasonable bounds. Instead, the inability to cope with the conflict more efficiently has made matters only worse.
Although much thought has been given to the dimensions of these kinds of conflicts and how to provide for a means of better managing them (Burgesses, Chapter Six) we feel that a common understanding of the nature of the problem by the various interest groups is crucial in establishing a sound basis for problem solving and negotiation. Decision Support Systems (DSS) can support establishing such a basis since they provide means to represent two types of information which are the best candidates for common acceptance: technical facts and mutual understanding of interests. In addition to these, DSS provide means to have different interests interact with this common base of technical facts, thus allowing stake-holders to invest in `possible worlds' (simulations of the real world) which allow changes and previews in time, rather than having to irreversibly risk `the real thing'.
In this paper we discuss the applicability of DSS technology in managing environmental conflicts such as the California drought. We will argue that since, as an environmental problem, the drought problem shares a set of commonalities with other multi-objective, multi-agent decision problems, DSS offer an attractive format for representing those problems, thus establishing a sound technical basis for problem analysis and the negotiation of solutions.
THE CONCEPT OF DSS - A BRIEF OVERVIEW
Figure 1. Integrated River Simulation System Decision Support Environment.
Figure 1 shows the conceptual (functional) structure of an environmental DSS (for more detailed overviews of DSS refer to Bosman 1983; Mittra, 1986; Sprague, 1986; Densham and Rushton, 1988; Reitsma 1990; Fedra, 1990). This conceptual structure reveals an integration of three basic components:
1) A modeling or simulation component;
2) A visualization and data-analysis component;
3) An evaluation, or decision-making component.
The modeling component offers a (simplified) representation of the various (physical) attributes of the decision problem. Here models for atmospheric transport, water supply and demand development, groundwater and surface water transport models, models that calculate risk and uncertainty contours, or models for urban development, industrial growth, or population forecasting reside. The function of these models is to simulate the consequences of possible management decisions for those physical parameters that need to be taken into account during the actual decision-making process (Loucks, et al., 1985). Note that this implies that not making decisions as to how to manage conflicts stemming from changes in supply, such as cases of drought or flooding, are considered decisions nonetheless. The very purpose of the modeling component of a DSS is to investigate the often complex relationships between policy and `the world the policy is imposed on', without having to actually try out that policy in real life first.
Figure 2. Municipal Water Districts.
Figure 3.
Figure 2, for instance, contains a display from a DSS developed by CADSWES for supporting water resources management. It is used to infer the likely effects of a particular scheme of groundwater pumping for groundwater salinity. The model generating these results is a dynamic model simulating the salinity of a groundwater aquifer over time as a function of a set of (pumping) stimuli exerted on that aquifer. Likewise, Figure 3, taken from a DSS developed by the International Institute for Applied Systems Analysis (IIASA), shows surface contours derived from running an atmospheric pollution model.
It should be noted that we consider initial conditions also a part of the modeling component. Although, by definition initial conditions form a static given (e.g., the original spatial distribution of reservoirs, water treatment plants, industrial, agricultural, or industrial water users), they serve an important purpose in DSS. Figure 3, for instance, combines the output of an atmospheric pollution transport model with static elements such as the locations of towns and different types of land-use. Such combinations support the evaluation of policy alternatives, since they allow inspection of the relationships between a policy alternative and its effects on other aspects that need to be taken into account in the decision making process.
Modeling does not always imply the use of more or less complex mathematical, physically-based models. Sometimes simply monitoring the development of an environmental system through a set of simple indicators may greatly improve the efficiency with which such a system can be utilized. Improving the efficiency is tantamount to increasing the `real' supply in the system thereby lowering the pressure on that system. A fine example of this type of DSS is a system developed by CADSWES for administering water rights on the Upper South Platte River in Colorado (Sieh and Eckhardt, forthcoming). Apart from a set of modeling tools which allow analysis of proposed water allocation schemes (`call structures'), the DSS processes real time data sent from stream gauges throughout the basin every fifteen minutes, relayed by satellite. That data structure, together with a data structure representing the topology of the basin and an algorithm for routing the water through that topology, constitutes a model of the basin. It represents both the static and dynamic components of the basin needed to simulate the effects of suggested call structures or call alterations. By only monitoring the river basin on smaller time intervals, however, the efficiency of water use in the basin can be significantly increased. On a fifteen-minute basis, for instance, individual precipitation events can now be detected by the stream gauge, and this additional supply is available for use. With larger time steps this is not the case. Empirically, water from precipitation events is available, of course, but this availability is merely theoretical. Since data on these events are not readily available in the case of a larger time step, they cannot be incorporated into the call-structure analysis. As a result, water users are told that they cannot divert water from the streams because of calculated shortages, while at the same time water from sudden precipitation events rushes by without being used.
Both Figure 2 and Figure 3 serve as an illustration of the second major component of a DSS: visualization and data-analysis. One of the main objectives of DSS is to visually represent the consequences of various policy alternatives. Therefore, models are integrated with more or less complex visualization tools which allow decision makers to `view' the expected consequences of proposed policy alternatives in a form that is most beneficial to their task. This graphic presentation of model results, as well as initial conditions, is tightly connected with data-analysis capabilities. In Figure 4, for example, a hydrologic model-output variable (outflow from a reservoir) is represented through a number of univariate statistics. These statistics may prove useful in the actual evaluation of the policy alternative which, coupled with a particular supply scenario, generated this outflow in the first place. Therefore, DSS often contain numerous visualization and data-analysis tools which provide the results from models in easy to understand, meaningful formats.
Figure 4.
As a `by-product' of both the modeling and visualization components of DSS, model validation can be continuously conducted. The integration of modeling and visualization within a DSS makes it possible to carry out large numbers of model runs within a limited amount of time. As such, testing models for accuracy and proper calibration becomes easier, and flaws in models are much more likely to be detected. An interesting issue associated with model validity and (organizational) decision making concerns the relation between model validity and the role a model plays in organizational decision making. Advocates of the so-called `organizational view of decision making' have argued that in organizational decision making organizational structure and processes, rather than rational decision making, are the driving forces (Keen and Morton, 1978; Huber, 1981). This is then reflected, for example, in the long-term use of a particular model or set of models during the decision-making process within the organization, without the actual validity of the model being questioned. Possible flaws in the model are exonerated in return for a well-established, well-structured decision process.
A third functional element of DSS is `evaluation'. Although this element is often implicitly represented in the other two components, it often provides a distinctly independent function in a DSS. This function is to formally evaluate model results (and thus policy alternatives!) with respect to various sets of decision criteria. Two of methods for doing so are often applied in DSS; implicit evaluation methods and explicit evaluation methods. In the case of implicit evaluation, information relevant for the decision making is merely displayed in one or more pictures, and it is the user's responsibility to `translate' this picture into some overall measure of attractiveness. Especially for very complex decision making situations involving many different interests represented by different interest groups or stake-holders, implicit evaluation methods may be very useful (Reitsma & Behrens, 1991). In implicit evaluation the evaluation and visualization/data-analysis components are strongly integrated.
Unlike implicit evaluation, explicit evaluation involves the application of specially developed (multi-criteria) evaluation models (Keeney and Raiffa, 1976; Wierzbicki, 1979; Nijkamp and Spronk, 1981; Grauer et al, 1982). These models apply mathematical constructs for combining a policy alternative's scores on various decision criteria into a single index often called `utility'. The models allow many different ways of combining 'part worth utilities' into overall utilities for the policy alternatives. Evaluation models also offer additional tools such as sensitivity analysis, mapping alternatives and criteria in multi-dimensional space, and evaluating alternatives relative to a desired distribution of criteria (Lewandowski and Grauer, 1982; Winkelbauer and Markstrom, 1990).
Yet another type of evaluation DSS may offer concerns optimization. As discussed by Johnson (Chapter Four) optimization consists of an empirical model with built-in explicit evaluation method. Through the application of a search algorithm, the DSS looks for system configurations which are optimal with respect to a pre-defined set of criteria. Note that this is a different approach than what was discussed before. In multi-criteria evaluation a fixed set of alternatives is evaluated on a fixed set of criteria. In the case of optimization, no pre-defined alternatives exist; the system is optimized on a pre-defined set of criteria. The resulting (optimized) solution is the only (possible) alternative.
In environmental decision making, however, optimization is often not a viable alternative. The reason for this resides with the multi-objective/multi agent nature of these problems. Multi-objective/multi-agent problems are characterized by very complex and often a priori unknown structure. As such, many optimization techniques cannot be applied since the problem is simply too rich, too complicated to be cast in the often simple `vocabulary' and `syntax' of optimization methods. Of course, parts of the problem can be optimized, for example, optimal release schedules for power generation. But the problem of environmental problem solving is not to find an optimal turbine scheduling in hydro-electric power plants. Instead, it is about bringing multiple objectives, represented by multiple agents together, in an effort to plan future actions such that those objectives are met or damaged at little as possible.
In a DSS, the three functional components are integrated into an easy-to-access, flexible-graphics computer environment. The most important advantage of such an integration is the creation of a platform that can be used by decision makers, managers, technical support staff, as well as anybody who has a definite interest in the problem at hand.
DSS AS A TECHNICAL BASIS FOR DISCUSSION AND NEGOTIATION
Environmental problem solving in modern societies often tends to get `messy'. Earlier we stated the main reasons for this: increasing pressure on environmental resources and increasing amounts of different interests involved in the conflict. As a result, conflicts often polarize and many opportunities for resolution get frustrated early on during the process (Burgesses, Chapter Six). DSS can help to prevent escalation and polarization from happening. In addition, the application of DSS technology can offer parties in environmental disputes a sound technical basis for discussion and negotiation.
Identical environmental resources may provide very different functions for different groups of actors. As a result, different interest groups attach different values to the various consequences of identical policy alternatives. Although, in itself this is quite natural and nothing much to worry about, it does create the problem of a communication barrier and difficulties in mutually understanding each other's points of view. This is not to be taken lightheartedly. Although one may be inclined to believe that a little bit of `good will' and some intelligence will be enough to make people understand each other's situation in case of a conflict in interest, an appropriate understanding of other people's or even one's own interests is sometimes difficult to acquire (Fisher and Ury, 1981; Reitsma, 1990).
A second cause of inefficient environmental problem solving concerns the information base on which partners in a conflict rely. Typically, interest groups have their own data-sources and rely on those when pursuing their objectives. As a result, information is only rarely shared, and information provided by opponents is considered selective and unreliable. As such, discussion and negotiation can be frustrated significantly. If different groups cannot or will not accept each other's information, positions are rigidly taken and communication and efforts to reach compromise solutions are severely hindered. Fisher and Ury (1981) refer to this situation as position-based negotiation versus interest-based negotiation.
By offering two related advantages, application of DSS can help solve these kinds of problems. First of all, since DSS are computer systems, DSS necessarily deal with the most formal aspects of a decision problem. Computers are not smart enough (yet) to be able to deal with emotions, feelings, preferences, etc. Instead they are used for the management of data with a very formal structure. This kind of data is often (though not exclusively) technical in nature. For environmental problems this means that DSS can at least deal with a lot of basic information representing the empirical characteristics of the problem. And since empirical data are in principal impartial, DSS databases and models could serve as a technical base for problem solution. This in its turn stresses the advantage of data-transparency which comes with DSS technology. Once information is stored in a DSS and once the procedures which operate on that data have been defined and accepted, the DSS must, by nature, be wholly objective. Since it has no interest and no capability to selectively manipulate data and models, it treats data always and ever in exactly the same manner. As such it constitutes a stable and reliable data source for problem solving.
A second advantage in using DSS in environmental problem-solving concerns the role the interaction with a DSS can play during the problem-solving process. We feel that when properly applied the very nature of the process of interacting with DSS offers some interesting opportunities for establishing efficient negotiation methods. In order to see this, however, elaboration is necessary on the nature of traditional single-user DSS versus recently developed, Group Decision Support Systems (GDSS).
SINGLE-USER DSS VERSUS GDSS
A problem which needs to be overcome in the application of DSS in complex environmental decision making concerns the ownership and accessibility of such a system. DSS have been developed for many purposes and for many different situations. Most systems have been constructed for single-actor decision making (e.g., Simonovic & Savic, 1989; Reitsma, 1990; Camara et al., 1990). By this we mean use by individuals or groups of individuals with very similar or even identical objectives. However, in many public policy questions, decisions are the result of a complicated social process in which various groups, each with its own goals and interests, try to influence the decision process to their benefit. This relates back to the problem of information-ownership mentioned earlier. If, for instance, one of the parties in a dispute develops a DSS for solving a problem, the likelihood that other opponent parties will be willing to accept that system and its associated data is not very high. In that context, most DSS seem to be rather alienated and esoteric pieces of automation technology. They are not what Kraemer once called the result of `social organization technology'. An interesting development in this respect is the construction of Group Decision Support Systems (DeSanctis & Gallupe, 1987; Kraemer & King, 1988; Pinsonneaut & Kraemer, 1989; Dickson, Scott Poole & DeSanctis, 1990; Zigurs, Poole and DeSanctis, 1988).
GDSS try to compromise between the inherently limited flexibility of information systems, and the complexity of the social decision making process in a group context. The idea behind these systems is that a GDSS provides ways to structure the process of decision making into a number of distinct tasks (e.g., agenda setting, criteria determination, negotiation, and voting) which lend themselves to various degrees of formalization. The set-up and operation of such a system allows full (personal) interaction between participants. Generally, these systems are used in the context of a `decision room' where the group of decision makers meets in an effort to solve a predefined problem. GDSS applies technology such as private monitors, access to shared models and databases, and public screens showing the overall status of the decision making process and each of its components. Since GDSS marries the advantages of DSS as a base for impartial, technical information with a social process, GDSS has some definite potential for establishing a sound basis for both data management and system interaction in environmental problem solving. However, matters are never as simple as that, not even in applied computer science and DSS technology.
SHARED MODELS AND VIEW-BASED APPROACHES
Although DSS and GDSS as described here offer substantial advantages for environmental decision making already, certain aspects need to be further developed in order to increase their potential even more. One approach is to blend the best of both traditional DSS and GDSS into a hybrid approach which we will denote `Shared-Model DSS'. In this case, the social interaction methods provided with a GDSS are augmented with the modeling capabilities of traditional DSS. The result are systems which have the technical bases of data and models for group consensus and group negotiation at their heart, but which allow multiple users to interact with those models, formulate different policy alternatives, and study model results. Although GDSS technology can then be used for implementing a structured system interaction and negotiation process, one more modification needs to be made: giving systems and users the capability to work with so-called `views'.
`Views' are pre-defined sets of data selections and data display formats for interpreting model results. Earlier, we mentioned implicit and explicit evaluation as a key component of DSS. It was also mentioned that, especially in decision making situations in which multiple actors and interests are involved, current explicit evaluation methods do not offer a sufficiently rich vocabulary for expressing the complexity of these situations. Views, therefore, apply implicit evaluation by allowing users of a shared-model DSS to define which information they want to monitor during various model runs, and how they want that information to be presented. The DSS remembers these definitions and monitors the models for the information the user requested. This allows users to very efficiently filter information from a DSS and evaluate that information with respect to their own criteria. Recently, proposals have been put forward for the development of a methodology which integrates classical model-based DSS, GDSS and view-based implicit evaluation into a group decision and negotiation context (Lewis, Reitsma, and Zigurs, 1991).
Another important aspect of group decision support systems is the ability of users to modify the system if some of its initial operating assumptions are deemed inappropriate. In case of single-user DSS, developers might suffice with carefully designed visualization tools developed in close cooperation with the sponsor. Obviously, if matters change and users decide that the existing visualizations do not fit their needs any longer, a new system design effort is needed.
In the case of GDSS, however, matters are more serious. Since GDSS are meant to function as platforms where representatives of multiple objectives can meet and work together for possible solutions of complex decision problems, the likelihood that developers can interact with all those possible user groups to determine their specific needs is rather low. Both the complexity of the problem and the dynamics of the environmental problem-solving process preclude such an approach. It is our opinion that for this kind of problem view-based (G)DSS offer an attractive solution. Views allow for strict formatting and formal approaches towards the selection and visualization of information from a model. On the other hand, those formalisms need not be determined a priori by the system developers, but can instead be decided upon by users while using the system.
TOWARDS DATA-DRIVEN SYSTEMS
Another set of recent developments in DSS for complex environmental problem-solving concerns practical drawbacks, often associated with the application of (large-scale) models in DSS. Currently, many models applied in DSS are large-scale models which require very specific data. Often, however, organizations which would like to use those models for future policy planning do not have all required data available. This seriously depreciates the value of the overall DSS, simply because such a DSS is model-based and is therefore to a large extent driven by running that model. This situation stems from the natural tendency in science and engineering to concentrate on general, rather then specific knowledge. As a result, large, generic models have been developed with substantial data requirements, making them very hard to apply in something like a DSS. As a result, emphasis is nowadays put on the data-driven nature of DSS.
In a data-driven approach DSS are constructed around available data. Models which can handle this data are either added to the system or constructed from scratch (Sieh and Eckhardt, forthcoming, and Reitsma and Sieh, forthcoming). This kind of a development offers interesting opportunities for the application of DSS in complex environmental decision making. For instance, carefully developing a DSS's databases from multiple sources is a process which helps participating groups understand each other's concerns and reasoning and helps them share insight into each other's information base. As such it supports the process of constructing a set of shared data, as well as models.
TOWARDS CASE-BASED SYSTEMS
Thus, with a growing emphasis on data rather than models, the domination of DSS by models is diminishing somewhat, although the arguments for shared model DSS remain valid. At the same time, however, developers are working on the construction of systems which apply case-based or analogy-based reasoning rather than models ((Gentner et al 1983, 1988; Keane, 1988; Thagard 1988; Kedar-Ceballi,1988; Kolodner et al 1985; Hammond, 1989). The thought behind those systems is both simple and powerful: `since people are good at learning from past experiences, we should try to develop systems which help people to capitalize on that capability.'
It should be noted that this constitutes an approach to decision making and problem solving which is very different from the ones we have discussed. Basically, development of case-based or analogy-based systems implies relying on large amounts of data on past conflict situations and smart reasoning schemes for drawing relationships and parallels between those cases and the case currently under dispute. No simulation model is applied here. In terms of more traditional DSS, case-based approaches rely almost exclusively on implicit evaluation methods. Users help systems to find analogous cases. Once presented with these analogies and perhaps some suggestions on what to learn from them, it is basically up to the user to `translate' the information on cases which are associated with the analogies onto their own `target' case. This heavy emphasis on implicit evaluation offers attractive opportunities for application in environmental DSS. Problems such as the California drought problem are amazingly complex. Therefore, explicit evaluation methods will be difficult to apply meaningfully and emphasis on implicit evaluation may be very worthwhile, including views and case-based approaches.
Although case-based technology is currently being developed, this technology is not readily available for application in environmental DSS. However, we expect some interesting impacts of this technology on the field of environmental DSS and on computer-supported decision making in general.
CONCLUSION
As the reader will have noticed, this chapter, unlike some of the other chapters, does not explicitly address the California drought problem. The reason for this is that DSS methodology implies a formal abstraction of the process of problem solving. To some extent, that abstraction is presented in Figure 1. This shows in a DSS context problem solving is separated into its three basic components: data and modeling, information display (visualization) and analysis, and evaluation. From such an abstract perspective, the differences between various kinds of environmental problems, regardless whether they concern droughts or floods, hazardous or radio-active waste management, land-use planning or power-plant siting, disappear. This abstraction of the problem-solving process into these common factors makes DSS a powerful technology. Abstraction implies the development of general methodologies which can be applied on many different situations. As such, DSS methodology forms the constant factor in DSS development, whereas the actual choice of data, models, system-interaction and visualization tools represent the differences in contents of different types of environmental problems.
Other chapters in this book clearly show that the California drought problem is a very complex problem. Although it may be far more complex than it actually needed to be, constructing a robust technological base for problem solving and negotiation requires the establishment and management of lots of data on the problem. This data, as well as sets of models which operate on that data, integrated into the framework of a DSS, may serve as a common base for stake-holders and negotiators searching for alternative solutions. Evaluative tools will show what interests are damaged under proposed alternatives and what possible compensations are available. Also, the sharing of information and the mutual management of that information contributes to mutual understanding and agreement on who's gaining and who's losing. From a political point of view, however, DSS will not make decision making any easier. The non-zero nature of many environmental problems simply implies that regardless of the particulars of a specific solution, some interests will gain or maintain their status, whereas others will have to give way. However, DSS can make the process more translucent and can force people to make explicit political decisions rather than decisions which are an intangible conglomerate of facts and value.
REFERENCES
Bosman, A. 1983. "Decision Support Systems. Problem Processing and Coordination." in Sol, H.G. (ed.). Processes and Tools for Decision Support. North Holland, pp.79-92.
Camara A.S. et al. 1990. "Decision Support Systems for Estuarine Water Quality Management." Journal of Water Resources Planning 116: 417-432.
Densham, P. & Rushton, G.R. 1988. "Decision Support Systems for Locational Planning" in Golledge, R.G. & Timmermans, H.J.P. (eds.). Behavioral Modeling in Geography and Planning. New York: Croom Helm.
DeSanctis, G. & Gallupe, R.B. 1987. "A Foundation for the Study of Group Decision Support Systems," Management Science 33:589-609.
Dickson, G.W., Scott Poole, M. & DeSanctis, G. 1990. An Overview of the Minnesota GDSS Research Project and the SAMM System. Minneapolis, MN: MIS Research Center, University of Minnesota.
Fedra, K. 1990. "Interactive Environmental Software: Integration, Simulation and Visualization." Laxenburg, Austria: IIASA RR-90-10; IIASA.
Fisher, R, & Ury, W. 1981. Getting to YES. Negotiating Agreement Without Giving In. New York: Penguin Books.
Genter, D., Falkenhainer, R. Skorstad, J. 1988 "Viewing Metaphor as Analogy." in Helman, D.H. (ed.). Analogical Reasoning: Perspectives of Artificial Intelligence, Cognitive Science, and Philosophy. Kluwer Academic Publishers, pp. 171-178
Grauer, M. Lewandowski, A. & Wierzbicki, A. 1982 "Methodology and Use of Multi-Objective Decision Making in Chemical Engineering." Laxenburg, Austria: IIASA RR82-37; IIASA.
Hammond, C. 1989. Case-Based Planning. Viewing Planning as a Memory Task. Academic Press.
Huber. 1981. "Nature of Organizational Decision Making and the Design of Decision Support Systems." MIS Quarterly.
Keane, M.T. 1988. Analogical Problem Solving. New York: Wiley and Sons.
Keen, P.G.W. & Scott Morton 1978. "Decision Support Systems: An Organizational Perspective." Addison-Wesley Series on Decision Support.
Kedar-Ceballi, S. 1988. "Analogy - From a Unified Perspective," in Helman, D.H. (ed.). Analogical Reasoning: Perspectives of Artificial Intelligence, Cognitive Science, and Philosophy. Kluwer Academic Publishers, pp. 65-104.
Keen, P.G.W. & Scott Morton, M.S. 1978. Decision Support Systems; An Organizational Perspective. Addison-Wesley Series on Decision Support.
Keeney, R.L. & Raiffa, H. 1976. Decisions with Multiple Objectives: Preferences and Trade-offs. New York: Wiley and Sons.
Kolodner, J., Simpson, R., Cyranski, K. 1985 "A Process Model of Case-based Reasoning in Problem Solving." Georgia Institute of Technology School of Information and Computer Science Technical Report 85/01.
Kraemer, K.L. & King, J.L. 1988. "Computer Based Systems for Cooperative Work and Group Decision Making," ACM Computing Surveys 20:115-146.
Lewandowski, A. & Grauer, M. 1982. The Reference Point Optimization Approach - Methods of Efficient Implementation. IIASA WP-82-26; IIASA, Laxenburg, Austria.
Lewis, C., Reitsma, R.F. Zigurs, I. 1990. "Model-based Decision Support: The Impact of Shared Simulation Models and Tailorable Information Viewing on Group Decision Making Outcomes and Processes"; NSF project grant proposal; CADSWES, University of Colorado at Boulder.
Loucks, D.P., Kindler, J. & Fedra, K. 1985. "Interactive Water Resources Modelling and Model Use: An Overview," Water Resources Research 21:95-102.
Mittra, S.S. 1986. Decision Support Systems: Tools and Techniques. New York: John Wiley & Sons.
Nijkamp, P. & Spronk, J. (eds.). 1981. Multiple Criteria Analysis: Operational Methods. Gower: Aldershot.
Pinsonneaut, A. & Kraemer, K. 1989. "The Impact of Technological Support on Groups: An Assessment of the Empirical Research," Decision Support Systems 5:197-216.
Reitsma, R.F 1990. "Functional Classification of Space; Aspects of Site Suitability Assessment in a Decision Support Environment," Laxenburg, Austria. IIASA RR-90-2.
Reitsma, R.F. & Behrens, J. 1991. "Integrated River Basin Management (IRBM): A Decision Support Approach," in: Klosterman, R. (ed.). Proceedings of the Second International Conference on Computers in Planning and Urban Management, Oxford July 6-8; pp.29-42.
Reitsma, R.F. and Sieh, D. (forthcoming). "Bootstrapping Models with Databases and GIS; Using Existing Data for Building Models and Decision Support Systems." Proceedings of the Annual Conference on Computing in Civil Engineering, Dallas, April 1992.
Simonovic, S.P. & Savic, D. 1989. "Intelligent Decision Support and Reservoir Management and Operations," Journal of Computing in Civil Engineering 3:367-385.
Sieh, D. & Eckhardt, J. 1991.
Sprague, R 1986. "A Framework for the Development of Decision Support Systems," in Sprague, R. & Watson, H.J. (eds.). Decision Support Systems; Putting Theory into Praxis. Englewood Cliffs, N.J.: Prentice Hall.
Thagard, P. 1988. "Dimensions of Analogy," in Helman, D.H. Analogical Reasoning: Perspectives of Artificial Intelligence, Cognitive Science, and Philosophy. Kluwer Academic Publishers, pp. 105-124.
Wierzbicky, A. 1979. "A Methodological Guide to Multi-Objective Optimization," Laxenburg, Austria: IIASA WP-79-122.
Winkelbauer, L. & Markstrom, S. 1990. "Symbolic and Numerical Methods in Hybrid Multi-Criteria Decision Support," Expert Systems with Applications 1:345-358.
Zigurs, I., Poole, M. S. and DeSanctis, G. 1988. "A Study of Influence in Computer-Mediated Group Decision Making." MIS Quarterly 12 (4): 625-644.