Simulation-based problem-solving environments for conflict studies

Simulation & Gaming http://sag.sagepub.com

Simulation-based problem-solving environments for conflict studies Levent Yilmaz, Tuncer I. Ören and Nasser Ghasem-Aghaee Simulation Gaming 2006; 37; 534 DOI: 10.1177/1046878106292537 The online version of this article can be found at: http://sag.sagepub.com/cgi/content/abstract/37/4/534

Published by: http://www.sagepublications.com

On behalf of: Association for Business Simulation & Experiential Learning International Simulation & Gaming Association Japan Association of Simulation & Gaming

North American Simulation & Gaming Association Society for Intercultural Education, Training, & Research

Additional services and information for Simulation & Gaming can be found at: Email Alerts: http://sag.sagepub.com/cgi/alerts Subscriptions: http://sag.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations http://sag.sagepub.com/cgi/content/refs/37/4/534

Downloaded from http://sag.sagepub.com by on October 13, 2009

Simulation-based problem-solving environments for conflict studies Levent Yilmaz Auburn University, United States Tuncer I. Ören University of Ottawa, Canada Nasser Ghasem-Aghaee University of Isfahan, Iran

At the dawn of the 21st century, uncertainty, change, and conflicts are inescapable facts of life. The challenge is to explore how the advances in simulation gaming as reflected in its state-of-the-art as well as in its potential can be helpful for conflict and peace studies. This article presents the issues, challenges, and foundations underlying multimodels and the multisimulation gaming strategy. Taxonomy of multimodels and plausible multisimulation realization strategies are presented to contribute to the development of advanced simulation-based problem-solving environments for social and political scientists to improve their ability to conceive, perceive, and foresee conflicting situations to ideally prevent them and—if they are inevitable—to resolve them. KEYWORDS: agents; conflict theory; gaming; multimodels; multisimulation; simulation methodology

Political, economic, ethnic, ideological, religious, military, as well as terrorist conflicts belong to the most destructive elements of the modern world. National and international conflicts are ubiquitous (Balencie & de la Grange, 1999). Conflict and peace studies include political, economic, ideological, and military conflict between nation-states or groups of nation-states, civil wars, and rebellions. In conflict studies, many activities other than combat are also involved. That is, conflict studies include political activities, diplomacy, economic warfare, as well as psychological operations. The 20th century was abundant with conflicts (Grant, 1992) and they are at least as worthy of study as during the post–Cold War period (Arquilla & Ronfeldt, 1997; Khalilzad & Lesser, 1998). Even more important, perhaps, is the study of conflict management (i.e., conflict avoidance and conflict resolution). Game theory that builds on the rationality principle has been applied to social problems (Geyer & van der Zouven, 1998; Shubik, 1964). However, evidence exists that classical game theory fails in cases where opponents have different value systems (Knight, Curtis, & Fogel, 1991; Lewis, 1999). Yet, different types of game theories (e.g., sequential games [Hamburger, 1979], repeated games [Banks & Sundaram, 1990; Leimar, 1997], differential games, evolutionary games [Lindgren, SIMULATION & GAMING, Vol. 37 No. 4, December 2006 534-556 DOI: 10.1177/1046878106292537 © 2006 Sage Publications

534 Downloaded from http://sag.sagepub.com by on October 13, 2009

Yilmaz et al. / SIMULATION-BASED PROBLEM SOLVING

535

1997], and hypergames [Fraser & Hipel, 1984]) as well as several other approaches such as bounded rationality (Simon, 1997), deterrence theory, and crisis destabilization continue to be used for their solutions. Conflicts, however, are complex social systems (Cheldelin, Druckman, & Fast, 2003; Schelling, 1960/1980). Novel simulation gaming formalisms, not conflicting with already proven theories and approaches, may be useful for the proper formulations and resolutions of conflicts. The premise of this article is based on the assumption that such methodologies can help us perceive, conceive, and foresee conflicting situations to ideally avoid them and—if they are inevitable—to resolve them. Regardless of their type and origin, conflicts are parts of social systems; similar to other social phenomena, they are difficult to model. Social systems are sometimes labeled in the literature as soft systems or ill-defined systems where the usefulness of traditional mathematical representations is questioned (Barnsley et al., 1998; Kaufmann, 1996; Spriet & Vansteenkiste, 1982; Waldrop, 1993). Powerful abstractions at the frontiers of computational modeling have been explored as well. Two such powerful abstractions include (a) the agent-based modeling paradigm (Conte, Sichman, & Gilbert, 1998; Ferber, 1999; Gilbert & Doran, 1998; Shoham, 1993; Tolk, 2005; Weiss, 1999), which provides a framework that enables computational specification of autonomy, adaptivity, and intelligence, and (b) exploratory modeling (Bankes, 1999; Davis, 2000, 2001; Davis & Bigelow, 1998, 2002; National Research Council, 1998), defined as the search or sampling over an ensemble of models that are plausible given a priori knowledge or are otherwise of interest (Bankes, 1993, 1994). Although these abstractions enable high-level, powerful, problem-solving capabilities, they need to be significantly enhanced to model realistic conflict phenomena. In conventional simulation, the results of a model run are viewed as a prediction of what we would expect to occur, which is contradictory to inherent uncertainty in human and social conflicts. We can view simulation as a laboratory, the results of which are interpreted as plausible outcomes of potentially unforeseen computational experiments that a scientist may not anticipate in advance. That is, the complete structure and parameters of computational experiments in conflict research may not effectively be decided in advance, as the trajectory of realistic conflicts is never fixed (Cheldelin et al., 2003; Kirchkamp, 1999). As Isard and Smith (1982) conclude, corresponding to the time path of structural change of a conflict situation should be a time path of proper conflict management procedures. By making recommendations for alternative models of such procedures, different modeling experiments can be produced. As any given experiment is based on a number of such recommendations, our knowledge about the problem being studied cannot be captured by any single game or experiment. Instead, the available knowledge is viewed as being contained in the collection of simulation experiments that become plausible and viable given what is known and learned during the simulation experiment. The central goal of this article is to lay out the rationale and conceptual foundations of multimodels and multisimulation methodology to significantly extend exploratory modeling in two dimensions to support conflict studies: (a) multisimulation with run-time updating capabilities to facilitate dealing with uncertainty and


536

SIMULATION & GAMING / December 2006

adaptivity in a single unified methodology and (b) run-time model recommendation, which monitors evolving processes, detects emergent situations and realities to make decisions for selection, staging, or switching of simulation components, thus aiding knowledge-based dynamic simulation composition at run-time.

Toward advanced methodologies for simulation-based peace and conflict studies Game theory (Binmore, 1994; Gibbons, 1997; Laszlo et al., 1977) and its extensions (Hofbauer & Sigmund, 1998; Liebrand & Messick, 1996) are widely used as models of social phenomena, including conflicts (Fraser & Hipel, 1984). Whereas seeking optimal solutions to specific social dilemmas is feasible by the use of analytic approaches, the limitations are extensively studied (Bennett, 1987). Computational methods for conflict theory and practice Social scientists are often interested in conducting experiments to explore alternative realities and assumptions to deal with uncertainty inherent in conflicts. The use of artificial intelligence (AI) tools, such as data mining and statistical machine learning, is pervasive in conflict avoidance and prevention (Fürnkranz, Petrak, Fahringer, Trappl, & Widmer, 2000; Trappl, Fürnkranz, & Petrak, 1996; Trappl, Fürnkranz, Petrak, & Bercovitch, 1997). Such methods often aim to predict the outbreak of conflicts by analyzing event data extracted from a multitude of sources (Stoll & Subramanian, 2002). Yet, we can use abilities of computers not only as passive databases that can only store, access, and reason about information but also as knowledge about the dynamics of conflict and peace processes to experiment with them under a multitude of scenarios. As such, simulation offers experimentation possibilities with “models” (that can represent our varied perceptions on reality) under a large number of scenarios, most of which would have lethal consequences if performed on the real system. In effect, simulation is goal-directed experimentation with dynamic models (i.e., models of systems with time-varying behavior and/or structure). Perceived from the proper perspective, simulation supports scientific thinking substantially. Experimentation is one of the key concepts in scientific thinking since Francis Bacon (1561-1626), who advocated it, over pure reasoning, in 1620 in his Novum Organum, “New Instrument,” in today’s parlance (Internet Encyclopedia of Philosophy: http://www.utm.edu/research/iep/b/bacon.htm). Bacon’s work was a categorical departure from and reaction to “Organon” (the Instrument), which was the title of logical works of Aristotle (384-322 B.C.) that had an “unparalleled influence on the history of Western thought” (Smith, 2001). Several types of modeling and simulation have been in use successfully for war gaming simulations (Ören, 2005). It is well known that peace studies are inherently more difficult than studies of war (Marom & Ben Itzak, 1999). The use of simulation gaming in international relations has benefited from the interaction between the foreign policy and defense communities. The early work on



537

the RAND Corporation’s POLEX simulations had a significant effect on the conceptualization of the inter-nation simulations (Guetzkow, 1995). The need for the simulation of actors within an international system led to tactical-decision simulations, where participants interact with a complex situation (such as a crisis), and social-system simulations, where participants interact with each other to drive the simulation forward. The success in enhancing the decision-making and negotiation skills of practitioners led to the educational use of simulation for nonpractitioners. Widely known simulation gaming exercises include MODEL UNITED NATION, MODEL ORGANIZATION OF AFRICAN UNITY, MODEL EUROPEAN UNION, as well as other models and simulations such as in international relations, INTERNATION SIMULATION (Guetzkow, Alger, Brody, Noel, & Snyder, 1963), GLOBUS, DIPLOMACY (Skidmore, 1993), NATIONS (Herzig & Skidmore, 1995), GLOBAL PROBLEMS (Lantis, 1998), and the INTERNATIONAL COMMUNICATION AND NEGOTIATION SIMULATIONS (ICONS). A more detailed discussion on the role of simulation in international relations education can be found in Starkey and Blake (2001). Kaufman (1998) argues for the value of using simulations as a tool for teaching international relations. Wolfe and Crookall (1998) emphasize the need for increased research on the educational virtues of simulation so that the method may be used more effectively. Vincent and Shepherd (1998) give an in-depth discussion of the research and education in simulation for international relations. Early contributions on the use of simulation and gaming in conflict analysis, management, and resolution community include Guetzkow and Valadez’s (1981) work that overviews simulation modeling of international processes, the work on simulation of negotiations (Saunders & Lewicki, 1995), conflict processes (Beriker & Druckman, 1996), political decision making (Druckman, Love, & Rozelle, 1983), and effectiveness of simulation-based conflict studies in education (Druckman, 1995). The significance of the validity of simulation models of conflict is also well recognized (Hermann, 1967; Sandole, 2003). There exists a separate line of simulation-based conflict analysis research within the political science field. The basis of such studies often involves game-theoretic analysis of the potential outcomes of dynamic interactions of opponents. Conventional games (Johnson, 1999; Liebrand & Messick, 1996) are concerned with what players ought to do in their own self-interest for optimized outcomes as opposed to what they might do. Agent-based social simulation (Conte & Castelfranchi, 1995) and the use of evolutionary computation (Holland, 1992; Kirchkamp, 1999), as exemplified in the work of Axelrod (1984), are noteworthy improvements in the use of simulation for social dilemmas (Gotts, Polhill, & Law, 2003). Yet, one of the criticisms against conventional forms of simulation in conflict analysis is that it is necessary to specify a specific structure on the game and its outcomes in advance so that a preference relationship for each player over the conflict can be established. This may be difficult to do when it is not obvious what is exactly involved in some of the outcomes (Yilmaz & Ören, 2004). Indeed, often in reality, the nature, parameters, and even the problem itself change during a conflict lifecycle (i.e., shifting from negotiation to mediation due to change in attitudes and action tendencies). Capabilities for


538


examining one’s options and exploring the conflict state space under alternative realities are important means to determine potential outcomes, especially when the utilities of options are not known a priori. In effect, corresponding to the time path of the structural change of a conflict situation should be a time path of the appropriate models of conflict management procedures (Isard & Smith, 1982). But, the question is, What should be the sequence of this shift pattern of models of conflict management procedures? Or should there be trigger mechanisms indicating when a shift should occur? Dynamic model and simulation updating The immediate factor raised by the above issues is that there is a need for runtime switching of models based on interpretation of emergent, potentially unforeseen conditions to facilitate dynamic run-time simulation composition and experimentation with multiple plausible models. A solution to this problem can also address the dynamic composability challenge raised in a recent RAND report (Davis & Anderson, 2003). Online model recommenders are needed to interface with the simulation kernel and the underlying operating environment to make recommendations as needed to explore the solution space intelligently. Exploration of the problem state space using a feasible sequence or stages of models would enable experimentation with alternative realities, potentially, at different levels of resolution. Detecting relevant and significant situations in a problem domain requires interpretation capabilities concerning emergent conditions and cause of observed effects. Observed effects need to be attributed to certain causes within the domain theory of the problem at hand. Such causes need to be appraised against the problemsolving goals and preferences to make recommendations for further, potentially simultaneous, exploration of different options. Whereas this scheme can be characterized as forward multisimulation, it is also worthwhile to examine the possibility of backtracking and replaying situated simulation histories with altered conditions and assumptions to explore alternative realities. The significance of dynamic model and simulation updating As the envelope of theory, methodology, and infrastructures increases, so does the perimeter of ambitious computational conflict research. An adaptive and exploratory multisimulation methodology is important at least for the following reasons: 1. In most realistic gaming scenarios, the nature of the problem changes as the simulation unfolds (Davis, 1988; Davis, Bigelow, & McEver, 2001). Initial parameters as well as scenarios can be irrelevant (i.e., real-time training scenarios) under emergent conditions. Relevant contingency models need to be identified and instantiated to continue exploration. 2. Our knowledge about the problem (i.e., conflict) being studied may not be captured by any single model or experiment. Instead, the available knowledge is viewed as being contained in the collection of all possible modeling experiments that are plausible



539

given what is known and what is learned. The power of a nation in conflict may depend on its economic performance, population activity, domestic resource exploitation, technology, military capability, political capacity, resilience, as well as the nature and level of sacrifice-readiness of the citizens, and interactions among the associated dynamics. 3. Dealing with uncertainty is paramount to analyzing complex evolving phenomena. Dynamic adaptivity in simulations and scenarios is necessary to deal with emergent conditions for evolving systems in a flexible manner. 4. As simulations of complex phenomena are used more and more to aid intuition, dynamic run-time simulation composition will help identify (re)solution strategies that are flexible, adaptive, and robust.

Multimodels Many real-world phenomena cannot be modeled by one single model; rather, they require the use of a set of complementary models that together are able to describe the whole process (i.e., conflict initiation, escalation, war, de-escalation, resolution, termination). A multimodel is a modular model that subsumes multiple submodels that together represent the behavior of the phenomena. Multimodel formalism was originally introduced by Ören (1987, 1991) to facilitate generalization of discontinuity in piecewise continuous systems. The formalism influenced the development of combined simulation, which entails the integration of continuous and discreteevent simulations within the same system description. For instance, as a special case of multimodel, coupled multiformalism specification developed by Praehofer (1992) extended DEVS (Zeigler, Praehofer, & Kim, 2000) formalism to provide a simulation environment for combined continuous/discrete-event simulation. Fishwick and Zeigler (1992) developed a specific type of multimodel to simulate qualitative dynamics of physical systems. Concomitantly, the EXTENSIBLE MODELING AND SIMULATION FRAMEWORK (XMSF; Blais et al., 2005) aims to facilitate dynamic extensibility next-generation simulation infrastructures. A taxonomy of multimodels: Exploring the design space of multimodels Social systems often have several aspects—some dormant—with mutual influence. To be able to study more than one aspect of reality and their interactions at the same time, one needs appropriate modeling methodologies such as multimodels. By expanding our horizon based on the variations on the multimodel structure and submodel activation behavior, we can identify various types of models. Table 1 illustrates a taxonomy of possible multimodel types based on various plausible constraints imposed on the submodel structure and activation policies. Based on the submodel structure of a multimodel, we consider the number of submodels active at a given time and the variability of the structure as the main criteria. Conventional multimodels, where only one model is active at a time, can be characterized as single-aspect (sequential) multimodels. In both cases, the states of the dormant models need to be saved and resumed once they are activated. Yet, re-instantiation of these latent models with new up-to-date state information is needed to facilitate Downloaded from http://sag.sagepub.com by on October 13, 2009

540


TABLE 1: Types of Multimodels

a. Synonyms are represented within parentheses.

continuity in the overall model behavior. Multiaspect models operate under conditions where more than one submodel can coexist simultaneously to represent distinct aspects of the same phenomena. Multimodels can have static or dynamic structures. Dynamic structure multimodels enable not only variation of the number of submodels but also their alteration (i.e., evolution). Extensible multimodels enable inclusion of new submodels that are unforeseen at the design time, whereas alteration of existing submodels results in evolutionary behavior or mutation of submodels. Different model update policies and mechanisms involving replacement, switching, and staging of submodels result in various design options for multimodels. The activation policy of a multimodel is based on the nature of information necessary for the activation of submodel(s) along with the location of this information. Depending on the



541

nature of the activation information, multimodels can be constraint driven, pattern directed, or goal directed. Constraint-driven multimodels are controlled by stationary (Fishwick & Zeigler, 1992) or adaptive transition policies that can have learning capabilities. Various reinforcement learning strategies can be used to maximize utilities, payoffs, or rewards in constraint-driven multimodel simulation. By choosing those actions in a given state that maximize a scalar reinforcement or feedback received after each action, the policy can adapt its action award structure by using a Markov decision process. Pattern-directed multimodels, on the other hand, follow a fixed cyclic or acyclic pattern (i.e., sequence) of submodel activations. Model selection can also be based on goal-directed (teleological) activation. Such models with planning facilities enable exploring a state space with alternative submodels, the preconditions of which are consistent with the observed conditions. In this viewpoint, submodels are simply operators in a state space search algorithm implemented by the planner that is interfaced with the multimodel. Pattern-directed activation can guide selection of known submodel(s). A simple case is metamorphic models where the number of submodels is finite and their sequence is fixed. Based on the location of information necessary for the activation of a submodel, we consider the following two cases. Active multimodels initiate updates due to internal emergent conditions that satisfy the transition conditions embedded with their specification (Ören, 1987). Such multimodels ideally incorporate a separate scheduler augmented with monitoring mechanisms to facilitate separation of concerns between model behavior generation and the behavior exhibited by the update protocols. In externally activated multimodels (i.e., passive multimodels), the decision to qualify and switch to a submodel is external to enable flexible and customizable update mechanisms or protocols.

Multiresolution, multiaspect, multistage multimodels of conflicts Multisimulation (or multisim, for short) is simulation of several aspects of reality in a study. It includes simulation with multiresolution models, simulation with multiaspect models, and simulation with multistage models. Simulation with multiresolution models involves building a single model, a family of models, or both to describe the same phenomenon at different levels of resolution in a mutually consistent way. Simulation with multiaspect models (or multiaspect simulation) allows computational experimentation with more than one aspect of reality simultaneously. This type of multisimulation is a novel way to perceive and experiment with several aspects of reality as well as exploring conditions affecting transitions between several aspects of systems. While exploring the transitions, one can also analyze the effects of encouraging and hindering transition conditions. Simulation with multistage models allows branching of a simulation study into several simulation studies, each branch allowing experimentation with a new model under similar or novel scenarios. Multisimulation is a simulation where at decision points, simulation updates may include decisions on


542


the branching of simulation studies as well as selection of models/submodels and associated parameters and experimental conditions to be used in subsequent stages of simulations. In multisimulation, at the interruption of the simulation, one can induce new model(s) based on the assessment of the situation so far. This very realistic possibility is not explored yet. Staging considers branching to other simulation studies in response to scenario or phase change during experimentation. In this section, we demonstrate the utility of multimodels in representing the structure of conflict situations as well as the associated dynamic behavior defined in terms of conflict management procedures. We use the framework and a concrete example proposed by Isard and Smith (1982) to substantiate our position on the utility of multimodels. Elementary concepts for modeling conflict situations A simple conflict situation is represented as CS = (Actors, Action, Environment, Outcome, Preferences). The actors component of CS stands for the participants in a conflict situation and includes, if necessary, a mediator (M). The set of actions, Actions, designates the acts that can be performed by the participants in conflict. Actions involve identifiable events that designate intentional change by a participant of his relation to the environment. In most conflict situations, proposals are involved to specify actions taken by each participant. Aside from the actions of participants, there exist environmental factors (e.g., weather, pollution level) that affect the decision of participants. These factors collectively make up the state of the environment. The combination of the environmental state and the action of participants produces an outcome, o ∈ Outcome.

Given the actions of other participants, the state of the environment, z, the outcome function, f(a, z) = o,

associates an outcome, o, with each action a that a particular participant may take. The Preference function maps the outcomes into a set of rankings. The preferences over actions are used during decision making to generate actions. This simple definition of a conflict situation allows us to distinguish between conflict and harmony. Harmony exists if the most preferred joint actions for each actor are the same. A conflict exists when there is no harmony joint action. A model of a conflict involves a conflict situation, which represents the structural components of the conflict, and a conflict management procedure (CMP) that represents the dynamic negotiation, arbitration, or mediation process used to resolve the conflict. A number of conflict management procedures are proposed for various types of conflicts (i.e., small number of options, large number of options, outcomes in relative terms) in Isard and Smith (1982).



543

Conflict model

i

Actorj

Information (Payoff, preferences)

i’

Actork

i+i’ z

Dj (decision making unit)

z Dk

CMP (via mediator) aj

ak

Oj (output/ production unit)

Ok

Ok

Oj Environment Architectural Representation of a Conflict Model FIGURE 1: Architectural Representation of a Conflict Model

Figure 1 illustrates the components and their relations with respect to each other in a conflict model. For the sake of simplicity, the number of actors is restricted to two. The CMP component facilitates interaction between participants of the conflict by realizing the proper conflict resolution mechanism. The mediator and the participants interact in accordance with the protocol designated by the CMP. The CMP component is where all the bargaining, negotiation, coalition, and strategic processes take place. The information subsystem embodies the payoffs, preferences, and utility information associated with the actions of participants. The decision unit of each actor ideally uses guiding principles such as cultural practices, as well as social and human factors such as personality and emotions to generate actions. In the absence of the encoding of such information, the payoffs and utilities associated with actions along with preferences over outcomes influence the decisionmaking process. This deliberation process is also influenced by the state of the environment. In our example, we will simply use the preferences over outcomes to select actions. The output/production unit associated with each actor computes the social, political, economic, ecologic, and technological implications of the action taken by the decision-making unit of the actor.


544


Level of K’s Program

Level of J’s Program

No Control (C0)

Limited Control (C1)

Moderate Control (C2)

Extensive Control (C3)

9, 9

3, 6

1, 3

2, 2

F

11, 10

8, 7

3, 4

5, 1

NP

12, 12

9, 11

6, 8

6, 4

AG

FIGURE 2: Joint Outcome Matrix

Case study: Strategic decision making via compromise over proposed outcomes/actions To aid exposition of the above ideas, we extend a case study originally developed by Isard and Smith (1982). In this case study, there are two participants. One of the participants (J) is the leader of a group of developed nations. The second (K) is a leader for a group of developing nations. The actions of the leader of the developed nations pertain to the type of development programs. These programs are as follows: 1. 2. 3. 4. 5.

No program [NP], Food [F], Agricultural Development and Food [AG], Industrial, Agricultural Development and Food [IND], Infrastructure, Industrial, Agricultural Development and Food [INF].

The actions of the developing nations are as follows: (a) No Control [C0], (b) Limited Control [C1], (c) Moderate Control [C2], (d) Extensive Control [C3], and (e) Full Control [C4]. The details for these programs are beyond the scope of this article and can be found in Isard and Smith (1982). Figure 2 presents the joint outcome matrix used by the mediator in managing a conflict. The contents of the cells in the matrix designate the payoff of actions for the participants. This matrix is stored within the information component of the conflict model and used to illustrate the utility of multiaspect and multiresolution modeling. Conflict management procedures Each conflict model is associated with a component that embodies the strategy used by the mediator to resolve a conflict situation. In this section, we consider two conflict management procedures: compromise over proposed outcomes and compromise over joint actions.



Level of B’s Program Level of A’s Program

No Control (C0)


Moderate Control (C2)

AG

80, 95

144, 125

168, 155

F

40, 70

112, 110

144, 150

0, 0

80, 50

120, 100

NP

545

FIGURE 3: Reduced Scale Matrix (Single Preferred Joint Outcome)

Compromise over proposed outcomes. Given a conflict state space as in the above example, consider the assumption that the full control option is unrealistic and excluded from the joint outcome matrix. This results in an information component that uses the matrix shown in Figure 3. This time, the matrix includes the preferences over the outcomes of both parties. The developing nations prefer to take the action designated by (AG, C2), whereas developed nations prefer (F, C3). Hence, we have a conflict that needs to be resolved. Within the simulation, the mediator having access to the complete outcome matrix points out and suggests the adoption of joint action (AG, C3), which yields a compromise outcome to each participant. The proposed joint action yields a better result compared to the outcome if both parties act on their own (i.e., [F, C2]). Action compromise principle. In the example shown in Figure 4, the mediator rules out the full control and infrastructure support options as infeasible actions. The preferences of individual participants are gathered by the mediator to establish the outcome matrix. A conflict appears as shown in Figure 4. The developed nations prefer (F, C3), whereas developing nations prefer (IND, C1). By asking each participant to concede a step, the mediator invokes the principle of compromise over the proposed joint actions. That is, the mediator contacts the developed nations’ leader to step up their program from F to AG, while demanding from the leader of the developing nations to step up their control from C1 to C2. The compromise joint action (AG, C2) is a comprise joint action. Next, we introduce the elementary concepts underlying multiresolution, multiaspect, and multistage models, followed by their use within conflict simulation within the proposed framework. Multiresolution, multiaspect, multistage multimodels of conflict situations Multiresolution modeling (MRM) is developing a single model, a family of models, or both to represent the behavior of a phenomenon at different levels of abstraction. The


546


Level of K’s Level Program of J’s Program

No Control (C0)

Limited Moderate Control (C1) Control (C2)

Extensive Control (C3)

IND

9, 13

1, 9

2, 7

3, 4

AG

13, 12

6, 8

3, 3

5, 2

F

15, 14

12, 9

6, 4

8, 1

NP

16, 16

13, 15

9, 11

9, 4

FIGURE 4: Multiple preferable joint actions

premise underlying MRM is that representation of the same phenomenon (i.e., conflict) at high-level and low-level resolution involves distinct yet dependent elements, constraints, and mechanisms. For instance, a strategic decision-making unit of an actor at a low-resolution model may have a small set of actions, compared with a high-resolution model with a large number of potential actions and detailed production processes (see next section). MRM deals with designing simulations that operate at different levels of resolution. MRM is based on model hierarchies, where the models lower in the hierarchy are at a higher level of resolution. Similarly, the abstraction of the overall conflict into multiple views, potentially overlapping views, is a critical issue in multiaspect modeling. Each view may designate a subset of the game, its action domain, and production mechanisms. For instance, whereas a model explores the effect of a set of actions on the political and economical domain, others may explore the ecological and technological aspects. Integration of distinct views becomes a critical issue in such types of multimodels. Multistage models are of significant interest, especially for conflicts. In a multistage model, the structures as well as the number of submodels are variable. Depending on the current phase of the problem, different families of models can be used to simulate the dynamics of a conflict. As the phase of a conflict shifts based on a number of factors such as the intensity of the conflict and associated threshold values, a new set of models would need to be used. The utility of multistage multimodels for conflict simulation In the compromise over the proposed outcome approach shown in Figure 3, there exists only one single joint action that is consistent with the principle. However, there can be several joint actions proposed by the mediator. For instance, in Figure 4, of all



547

joint actions consistent with the compromise outcome principle, the developed nations prefer the joint action (AG, C3), whereas the developing nations prefer (IND, C2). In this case, one can branch the conflict simulation to observe the effects of choosing alternative joint actions that influence the production unit, which in turn affects the environment of the conflict. Figure 5 demonstrates how a multimodel can be used to explore both branches based on alternative decisions. In this scenario, whereas CMA carries out the simulation of the production processes based on the joint action (AG, C3), CMB explores the effect of the production processes based on the (IND, C2) joint action. There are conflict situations in which sequential staging of submodels can be useful as well. Consider, for instance, the case in Figure 4, when a first try at reaching a compromise solution has not worked. The mediator may request that the participants focus on the next preferred joint action. In this case, the developing nations will propose (IND, C2), whereas the developed nations propose (AG, C3). Because the joint actions are still incompatible, the mediator may now suggest using a new conflict management procedure (i.e., compromise outcome principle). The use of this CMP results in two plausible joint actions: (AG, C2) and (IND, C3). The mediator may insist on one option or branching may take place. Yet, it is clear that a change in the conflict management procedure indicates staging to a different submodel that embodies the compromise outcome strategy as part of the CMP subsystem of the conflict model. In effect, the simulation starts with the active baseline submodel that uses compromise over joint actions strategy and then after two rounds of interactions, the submodel that uses the compromise over outcomes strategy becomes active due to switching to a new strategy. The utility of multiaspect multimodels for conflict simulation With the actions of two participants well defined, Figure 2 presented a payoff matrix used by the decision unit of the actors in the conflict model. The numbers depict the preferences over the perceived outcomes. If the level of J’s (leader of developed nations) program is NP, whereas the leader of the developing nation chooses no control (C0), then both participants realize zero payoff. When the leader of developing nations chooses the no control option, the shift by the developed nations from any level of program to a higher level option yields a perceived payoff increase of 40 by the developing nation. On the other hand, if limited control is chosen by the developing nations, the payoff will increase to 80. As developed nations move by step from any level to a high-level program, the developing nations would perceive the increase in their payoff to be only 32. By changing the payoff values and increments in the matrix, various strategies can be tested with distinct submodels. For instance, the increment in the perceived payoff in the case of limited control is less than if the developing nations were to have chosen no control. This is mainly due to more organization and less assistance in their economies than if they allowed riots under C0. Decomposing a complex joint action matrix into distinct (or potentially overlapping) submatrices or changing the conflict management procedure within a conflict model results in a distinct set of behaviors based on the observed state of the environment and interactions with other


548


CMPB

P’0

D’0

PA

DA

ENV

CMPA

PMA

P’A

D’A

CMA (Active)

PB

DB CMPB

PMB

CMB

P’0

D’0

DA

P’B

D’B

PA

FIGURE 5: Exploratory Simulation With Multistage Multimodels

P0

D0

PMB

CM0

P0

CMPB

ENV

CMPA

PMA

PMB

D0

CMA

CM0 (Active)

P’A

D’A

P0

D0

CMPB

CMPB

PMB

CM0

PB

DB P’B

D’B

P’0

D’0

PMB

CMB

PA

DA

ENV

CMPA

PMA

P’A

D’A

CMA (Active)

PB

DB

CMPB

PMB

P’B

D’B

CMB (Active)


549

Multiaspect Multimodel

DA

CMA

CMB

CMC

PMA

PMA

PMC

CMP

A

PA

D’A

DB

P’A

PB

ENV’

CMP

A

D’B

DA

P’B

PC

ENV’

CMP

A

D’A P’C

ENV’

View Integration

ENV

Conflict Model (CM) Payoff Matrix (PM) FIGURE 6: Multiaspect Multimodel

participants. Each such conflict model may capture a specific aspect of the problem domain by varying the associated component and encapsulating them within distinct submodels. Note, however, that the results of distinct views need to be integrated to facilitate derivation of a coherent environment (see Figure 6). Multiaspect multimodels will prove useful, in particular, when the overall required production components of a model cannot be captured within a single conflict model due to extensive complexity and/or discrepancy underlying the requisite formalism (i.e., continuous vs. discrete-event) needed to capture the behavior of the production system. In such cases, alternative production (output) models such as macro-models of social group production, econometric models, ecological models stemming from economic activity, and so on can be encapsulated in distinct submodels that are simulated simultaneously. For details on plausible design strategies in developing such models, we refer interested readers to Yilmaz and Oren (2004). The utility of multiresolution multimodels for conflict simulation Multiresolution multimodels are useful in at least two cases in conflict simulation within the context of the above conflict modeling framework: (a) Whereas decision


550


makers are interested in a high-level view of the conflict with a small number of available options and actions, strategists need to drill-down to a high-resolution view of the conflict, which requires detailed models with large number of options and joint actions. Simultaneous simulation of a conflict situation at different levels of resolution with a potential flow of information for calibration (i.e., via environment component) of models improves the credibility and fidelity of conflict simulation. (b) As a conflict simulation unfolds, the available actions and the associated payoff matrix used by the actors within models need to updated, because the conflict situation evolves. When the CMPs used with situations with a small number of options do not scale and are not effective for situations with a large number of options, new CMPs need to be used. This requires multiresolution models with multiple submodels that encapsulate different payoff matrices in conjunction with proper CMPs. Figure 7 illustrates the case in which the simulation starts with a limited payoff matrix with only three viable options for each leader. If the leaders of the developed nations and developing nations choose the actions AG and C2, respectively, then the multiresolution model switches to the new model depicted as CMB. This new model uses not only a new payoff matrix within its information subsystem but also a proper CMP (i.e., split the difference; Isard & Smith, 1982) that is designed to operate on a large number of options. Note that both CMA and CMB are active at the same time. The changes induced by the high-resolution model need to be reflected to the decisionmaking unit of the CMA for calibration of the low-resolution model. On the need for multimodels for conflict simulation Some of the concrete examples discussed above emphasized the importance of dropping the notion of using a single conflict management procedure for the management and resolution of complex conflicts. The discussion on the use of multiaspect, multistage, multiresolution multimodels implies a certain type of conflict dynamics, that is, a set of stages in the process associated with proper conflict management procedures for each stage. Note, however, that as conflict simulation unfolds, the parameters of the decision and payoff matrices, the state space of the problem, the attitudes, and preferences may also change. Therefore, the time path of a conflict should map onto a time path of conflict resolution strategies embedded within the models. Critical questions that need to be answered include the issues pertaining to the mechanism by which resolution techniques are selected, when and how shifts occur in updating multimodels, and to whom the judgment to determine the shift should be given.

Conclusions Novel modeling and simulation formalisms, not conflicting with already proven theories and approaches, may be useful for the proper formulations of conflicts (Ören, 2001). Such methodologies can help us perceive, conceive, and foresee conflicting situations to ideally avoid them and—if they are inevitable—to resolve them



551

Multiresolution Multimodel CMA

CMB

PMA DA

CMPA

PA

PMB D’A

DB

P’A

PB

CMPB

D’B

P’B

ENV CMPA: Compromise over proposed outcomes CMPB Split the difference

PMA Level of B’s Level Program of A’s Program

No Control (CO)



AG

80, 95

144, 125

144, 125

F

40, 70

112, 110

112, 110

0,0

80, 50

80, 50

NP

PMB Level of K’s Limited Moderate Extensive Full Level Program No Control (CO) Control (C1) Control (C2) Control (C3) Control (C4) of J’s Program INF

160, 40

208, 50

216, 60

184, 70

122, 80

IND

120, 90

176, 110

192, 130

168, 150

114, 170

AG

80, 95

144, 125

168, 155

152. 185

106, 215

F

40, 70

112, 110

144, 150

136, 190

98, 230

NP

0, 0

80, 50

120, 100

120, 150

90, 200

FIGURE 7: Multiresolution Multimodel

(Yilmaz & Ören, 2004). In classical mechanics, it is customary to study one aspect of reality at a time. To model such uncertain, evolving, and multistaged social phenomena, we need to explore and extend existing powerful abstractions at the frontiers of Downloaded from http://sag.sagepub.com by on October 13, 2009

552


computational modeling. Two such powerful abstractions include (a) the agentbased modeling paradigm that provides a framework that enables computational specification of autonomy and intelligence and (b) exploratory modeling (Bankes, 1993), defined as the search or sampling over of an ensemble of models that are plausible given a priori knowledge or are otherwise of interest to deal with uncertainty. Although these abstractions enable high-level and powerful problem-solving capabilities, they need to be significantly enhanced to realize solutions for complex conflicts, including uncertainty, as well as the cases where the nature of the conflict changes as the simulation unfolds, and information about the actual conflict is acquired during simulation rather than before. To this end, the conceptual foundations of a new exploratory multisimulation methodology with dynamic model and simulation updating are presented as an advanced problem-solving environment for social and political scientists to observe and examine the implications and plausible outcomes of decisions in conflict. Strategies and options for the realization of multimodels and multisimulation are overviewed to facilitate the development of nextgeneration problem-solving environments for conflict studies.

References Arquilla, J., & Ronfeldt, D. (1997). A new epoch and spectrum of conflict. Santa Monica, CA: RAND. Retrieved from http://www.rand.org/publications/MR/MR880/ Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books. Balencie, J–M., & de la Grange, A. (1999). Guerres civiles et violences politiques [Rebellious world: Civil wars and political violences]. In Encyclopedie des conflits [Encyclopedia of the conflicts]. Paris: Editions Michalon. Bankes, S. (1993). Exploratory modeling for policy analysis. Operations Research, 4(3), 1993. Bankes, S. (1994). Exploring the foundations of artificial societies: Experiments in evolving solutions to iterated N-player prisoner’s dilemma. In R. A. Brooks & P. Maes (Eds.), Artificial Life IV (pp. 337–342). Cambridge, MA: MIT Press. Bankes, S. (1999). Tools and techniques for developing policies for complex and uncertain systems. Proceedings of the National Academy of Sciences, Colloquium, 99(3), 7263-7266. Banks, J. S., & Sundaram, R. K. (1990). Repeated games, finite automata, and complexity. Games and Economic Behavior, 2, 97–117. Barnsley, M. F., et al. (1988). The science of fractal images. New York: Springer-Verlag. Bennett, G. P. (1987). Analyzing conflict and its resolution: Some mathematical contributions. Oxford: Clarendon Press. Beriker, N., & Druckman, D. (1996). Simulating the Lausanne peace negotiations, 1922-23: Power asymmetries in bargaining. Simulation & Gaming, 27(2), 162-183. Binmore, K. G. (1994). Game theory and the social contract—Volume 1: Playing fair. Cambridge, MA: MIT Press. Blais, C., Brutzman, D., Drake, D., Moen, D., Morse, K., Pullen, M., & Tolk, A. (2005). Extensible modeling and simulation framework (XMSF): 2004 Project Summary Report (NPS-MV-05-002). Monterey: Naval Postgraduate School MOVES Institute. Cheldelin, S., Druckman, D., & Fast, L. (2003). Conflict: From analysis to intervention. New York: Continuum. Conte, R., & Castelfranchi, C. (1995). Cognitive and social action. London: UCL Press. Conte, R., Sichman, J. S., & Gilbert, N. (1998). MAS and social simulation: A suitable commitment. In J. S. Sichman, R. Conte, & N. Gilbert (Eds.), Multi-agent systems and agent-based simulation: First international workshop MABS ‘98 (pp. 1–9). Berlin: Springer. Downloaded from http://sag.sagepub.com by on October 13, 2009


553

Davis, K. P. (1988). The role of uncertainty in assessing the NATO-PACT central region balance (RAND N-2839). Santa Monica, CA: RAND. (Also available from General Accounting Office, NATOWarsaw Pact Conventional Force Balance, GAO/NSIAD-89-23B.) Davis, K. P. (2000). Exploratory analysis enabled by multiresolution, multiperspective modeling. In A. Jeffrey, R. R. Joines, K. K. Barton, & P. Fishwick (Eds.), Proceedings of the 2000 Winter Simulation Conference (RP-925). Santa Monica, CA: RAND. Davis, K. P. (2001). Analytical architecture for capabilities-based planning: Mission-system analysis, and transformation. Santa Monica, CA: RAND. Davis, K. P., & Anderson, A. R. (2003). Improving the composability of department of defense models and simulations [RAND Monograph]. Santa Monica, CA: RAND. Davis, K. P., & Bigelow, J. (1998). Experiments in multiresolution modeling (RAND MR-1004). Santa Monica, CA: RAND. Retrieved from http://www.rand.org/publications/MR/MR1004/ Davis, K. P., & Bigelow, J. H. (2002). Motivated metamodels: Synthesis of cause-effect reasoning and statistical modeling. Santa Monica, CA: RAND. Davis, K. P., Bigelow, J., & McEver, J. (2001). Exploratory analysis and a case history of multiscenario, multiperspective modeling (RAND RP-925). Santa Monica, CA: RAND. Druckman, D. (1995). The educational effectiveness of interactive games. In D. Crookall & K. Arai (Eds.), Simulation/gaming across disciplines and cultures: ISAGA at a watershed (pp. 178-187). Thousand Oaks, CA: Sage. Druckman, D., Love, R., & Rozelle, R. (1983). Resolving conflicts of interest and ideologies: A simulation of political decision-making. Social Behavior and Personality: An International Journal, 11, 23-28. Ferber, J. (1999). Multi-agent systems: An introduction to distributed artificial intelligence. Harlow, UK: Addison Wesley. Fishwick, P., & Zeigler, P. B. (1992). A multimodel methodology for qualitative model engineering. ACM Transactions on Modeling and Simulation, 2(1), 52-81. Fraser, M. N., & Hipel, W. K. (1984). Conflict analysis models and resolutions. New York: Elsevier Science Publishers. Fürnkranz, K. K., Petrak, J. J., Fahringer, P. B., Trappl, R., & Widmer, G. (2000). Searching for patterns in political event sequences: Experiments with the KEDS database. Cybernetics and Systems, 31(6), 649-668. Geyer, F., & van der Zouven, J. (1998). Bibliography on sociocybernetics (3rd ed.). Retrieved from http://www.unizar.es/sociocybernetics/bibliografia.html Gibbons, R. (1997). An introduction to applicable game theory. Journal of Economic Perspectives, 11(1), 127–149. Gilbert, N., & Doran, E. J. (1998). Social science microsimulation. New York: Springer. Gotts, M. N., Polhill, G. J., & Law, R.N.A. (2003). Agent-based simulation in the study of social dilemmas. Artificial Intelligence Review, 19, 3-92. Grant, N. (1992). Illustrated history of 20th century conflict. London: Reed Consumer Books. Guetzkow, H. (1995). Recollections about the inter-nation simulation (INS) and some derivatives in global modeling. Simulation & Gaming: An International Journal, 26(4), 453-470. Guetzkow, H., Alger, F. C., Brody, A. R., Noel, C. R., & Snyder, C. R. (1963). Simulation in international relations: Developments for teaching and research. Englewood Cliffs, NJ: Prentice Hall. Guetzkow, H., & Valadez, J. J. (1981). Simulation and reality: Validity research. In H. Guetzkow & J. J. Valadez (Eds.), Simulated international processes: Theories and research in global modeling (pp. 253-330). Beverly Hills, CA: Sage. Hamburger, H. (1979). Games as models of social phenomena. San Francisco: W. H. Freeman. Hermann, C. F. (1967). Validation problems in games and simulations with special reference to models of international politics. Behavioral Science, 2(3), 216-231. Herzig, M., & Skidmore, D. (1995). Nations: A simulation game in international politics. In Pew case studies in international affairs (No. 169). Washington, DC: The Institute for the Study of Diplomacy. Hofbauer, J., & Sigmund, K. (1998). Evolutionary games and population dynamics. New York: Cambridge University Press.


554


Holland, J. H. (1992). Adaptation in natural and artificial systems (2nd ed.). Cambridge, MA: MIT Press. Isard, W., & Smith, C. (1982). Conflict analysis and practical conflict management procedures: An introduction to peace science. Cambridge, MA: Ballinger. Johnson, P. E. (1999). Simulation modelling in political science. American Behavioral Scientist, 42(10), 1509–1530. Kaufman, J. (1998). Using simulation as a tool to teach about international negotiation. International Negotiation: A Journal of Theory and Practice, 3(1), 59-75. Kaufmann, S. (1996). At home in the universe: The search for laws of self-organization and complexity. Oxford, UK: Oxford University Press. Khalilzad, Z., & Lesser, I. O. (1998). Sources of conflict in the 21st century—Regional futures and U.S. strategy. Santa Monica, CA: RAND. Kirchkamp, O. (1999). Simultaneous evolution of learning rules and strategies. Journal of Economic Behavior and Organization, 40, 295–312. Knight, D. E., Curtis, W. H., & Fogel, L. J. (Eds.). (1991). Cybernetics, simulation, and conflict resolution, Proceedings of the 3rd annual symposium of the American Society for Cybernetics. New York: Spartan Books. Lantis, J. S. (1998). Simulations and experiential learning in the international relations classroom. International Negotiation: A Journal of Theory and Practice, 3(1), 39-57. Laszlo, E., et al. (1977). Goals for mankind. New York: E. P. Dutton. Leimar, O. (1997). Repeated games: A state space approach. Journal of Theoretical Biology, 184, 471–498. Lewis, R. D. (1999). When cultures collide—Managing successfully across cultures. London: Nicholas Brealey. Liebrand, W.B.G., & Messick, D. M. (1996). Computer simulations of sustainable cooperation in social dilemmas. In R. Hegselmann, U. Mueller, & K. G. Troitzsch (Eds.), Modelling and simulation in the social sciences from the philosophy of science point of view (pp. 235–247). Berlin: Springer-Verlag. Lindgren, K. (1997). Evolutionary dynamics in game-theoretic models. In B. Arthur, S. Durlauf, & D. Lane (Eds.), The economy as an evolving complex system II (pp. 337-367). Reading, MA: Addison-Wesley. Marom, S., & Ben Itzak, U. (1999). Training for peace-keeping operations—Unfamiliar challenges. Modern Simulation & Training, 2, 46-49. National Research Council. (1998). Modeling and simulation technology for the United States Navy and Marine Corps, 2000-2035 (Vol. 9). Washington, DC: National Academy Press. Ören, T. I. (1987, July 27-30). Model update: A model specification formalism with a generalized view of discontinuity. In Proceedings of the summer computer simulation conference, Montreal, Quebec, Canada (pp. 689-694). San Diego, CA: SCS Press. Ören, T. I. (1991). Dynamic templates and semantic rules for simulation advisors and certifiers. In A. P. Fishwick & B. R. Modjeski (Eds.), Knowledge-based simulation: Methodology and application (pp. 53-76). New York: Springer-Verlag. Ören, T. I. (2001). Towards a modelling formalism for conflict management. In S. H. Sarjoughian & E. F. Cellier (Eds.), Discrete-event modeling and simulation: A tapestry of systems and AI-based theories and methodologies (pp. 93-106). New York: Springer-Verlag. Ören, T. I. (2005, November 28-December 1). Toward the body of knowledge of modeling and simulation (M&SBOK). In Proceedings of I/ITSEC (Interservice/Industry Training, Simulation Conference), Orlando, Florida [Invited Tutorial]. Arlington, VA: NTSA. Praehofer, H. (1992). System theoretic foundations for combined discrete-continuous system simulation. Unpublished doctoral dissertation, Johannes Kepler University, Linz, Austria. Sandole, D.J.D. (2003). Validating simulation-based models of conflict. Simulation & Gaming, 34(2), 249-280. Saunders, D. M., & Lewicki, J. R. (1995). Teaching negotiation with computer simulations: Pedagogical and practical considerations. Negotiation Journal, 11(2), 157-167. Schelling, T. (1980). The strategy of conflict (Reprint ed.). Cambridge, MA: Harvard University Press. (Original work published 1960) Shoham, Y. (1993). Agent oriented programming. Artificial Intelligence, 60, 51-92. Shubik, M. (Ed.). (1964). Game theory and related approaches to social behavior. New York: Wiley.



555

Simon, H. A. (1997). Models of bounded rationality—Volume 3: Empirically grounded economic reason. Cambridge, MA: MIT Press. Skidmore, D. (1993). Group projects and teaching international relations. International Studies Notes, 18(3), 49-53. Smith, R. (2001, Winter). Aristotle’s logic. In Stanford encyclopedia of philosophy. Retrieved from http://plato.stanford.edu/archives/win2001/entries/aristotle-logic/ Spriet, J. A., & Vansteenkiste, G. C. (1982). Computer-aided modelling and simulation. London: Academic Press. Starkey, A. B., & Blake, L. E. (2001). Simulation in international relations education. Simulation & Gaming, 32(4), 537-551. Stoll, R., & Subramanian, D. (2002). Events, patterns, and analysis: Forecasting international conflict in the twenty-first century. In R. Trappl (Ed.), Computer-aided methods for international conflict resolution and prevention (pp. 145-160). New York: Springer-Verlag. Tolk, A. (2005). An agent-based decision support system architecture for the military domain. In G. E. PhillipsWren & L. C. Jain (Eds.), Intelligent decision support systems in agent-mediated environments, Volume 115, Frontiers in artificial intelligence and applications (pp. 187-205). Amsterdam: IOS Press. Trappl, R., Fürnkranz, J., & Petrak, J. (1996). Digging for peace: Using machine learning methods for assessing international conflict databases. In Proceedings of the 12th European conference on artificial intelligence (ECAI-96), Budapest, Hungary (pp. 453-457). New York: John Wiley & Sons. Trappl, R., Fürnkranz, J., Petrak, J., & Bercovitch, J. (1997). Machine learning and case-based reasoning: Their potential role in preventing the outbreak of war or in ending them. In R. G. Della, R. Kruse, & J-H. Lenz (Eds.), Learning, networks and statistics: CISM courses and lectures (No. 382, pp. 209-225). New York: Springer-Verlag. Vincent, A., & Shepherd, J. (1998). Experiences in teaching Middle East politics via Internet-based roleplay simulations. Journal of Interactive Media in Education, 98(11), 23-37. Waldrop, M. M. (1993). Complexity: The emerging science at the edge of order and chaos. New York: Touchstone Books. Weiss, G. (1999). Multiagent systems: A modern approach to distributed artificial intelligence. Cambridge, MA: MIT Press. Wolfe, J., & Crookall, D. (1998). Developing a scientific knowledge of simulation/gaming. Simulation & Gaming, 29, 7-19. Yilmaz, L., & Ören, T. (2004). Dynamic model updating in simulation with multimodels: A taxonomy and generic agent-based architecture. In Proceedings of the SCSC’ 04 (pp. 3-8). San Diego, CA: SCS Press. Zeigler, P. B., Praehofer, H. P., & Kim, G. T. (2000). Theory of modeling and simulation: Integrating discrete event and continuous complex dynamic systems. San Diego, CA: Academic Press.

Levent Yilmaz is an assistant professor of computer science and engineering in the College of Engineering at Auburn University and cofounder of the Auburn Modeling and Simulation Laboratory of the M&SNet. He received his PhD and MS degrees from Virginia Polytechnic Institute and State University. His research interests are in advancing the theory and methodology of simulation modeling, agent-directed simulation (to explore dynamics of sociotechnical systems, organizations, and human/team behavior), and education in simulation modeling. He is a member of ACM, IEEE Computer Society, Society for Computer Simulation International, and Upsilon Pi Epsilon. Tuncer I. Ören is a professor emeritus of computer science at the School of Information Technology and Engineering of the University of Ottawa, Canada. He has contributed to (a) the advancement of modeling and simulation methodologies; (b) the synergy of simulation, system theories, cybernetics, artificial intelligence, and software engineering; (c) reliability issues in modeling and simulation; and (d) ethics in simulation. He is currently a member of the board of directors, AVP on Ethics, and the director of the McLeod Modeling and Simulation Network of the Society for Modeling and Simulation International. He has completed more than 330 publications, some translated in Chinese and German. He has contributed


556


to more than 300 conferences and seminars held in 28 countries; has received invitations from the United Nations, sponsorship from NATO, and fellowships or sponsorships in several countries; and has more than 20 Who’s Who citations. He has received the Information Age Award from the Turkish Ministry of Culture (1991) and plaques and certificates of appreciation from organizations including NATO, Atomic Energy of Canada, and ACM. Nasser-Ghasem Aghaee is an associate professor of computer engineering at the Faculty of Engineering of the University of Isfahan. He earned his PhD and MS degrees from the University of Bradford and Georgia Tech, respectively. His research interests are in computer simulation object-oriented analysis and design, artificial intelligence and expert systems, AI in software engineering, and AI in simulation. ADDRESSES: LY: 107 Dunstan Hall, M&SNet: Auburn Modeling and Simulation Laboratory, Department of Computer Science and Software Engineering, College of Engineering, Auburn University, Auburn, AL 36849, USA; telephone: 001-334-844-6343; fax: 001334-844-6329; e-mail: [email protected]. TIÖ: M&SNet: Ottawa Center of the McLeod Institute of Sim. Sciences, School of Information Technology and Engineering (SITE), University of Ottawa, 800 King Edward, Room 4068, P.O. Box 450, Stn A, Ottawa, Ontario, Canada, K1N 6N5; telephone: +1 613-562-5800, ext. 2162; fax: +1 613-562-5664; e-mail: [email protected]. NGA: Department of Computer Engineering, Faculty of Engineering, University of Isfahan, Daneshgah Avenue, Isfahan, Postal Code 81746-73441, Iran; telephone: +98-311-7713810; fax: +98-311-6682887.


Simulation-based problem-solving environments for conflict studies

Simulation-based problem-solving environments for conflict studies

Suggest Documents