A GROUP DECISION SUPPORT SYSTEM

A description on models and modules in GDSS based on cooperative MAS

Shi-Weiren, Jiang-Daoping, Liang-Yonglin, Chen-Jing

College of Automation, Chongqing University, Chongqing, China

Key word: GDSS, Multi-Agent, Cooperation, Negotiation

Abstract: GDSS is a popular and attractive topic in decision f

ield. It is reasonable to introduce multi-Agent

technology and methodology into GDSS because they are both distributed systems and support interaction

in group members. We look at the development of GDSS as being a process of putting together a

coordinated workflow of collaborating Agents that is able to support a problem-solving process. We

propose models to describe character of Agents and issue in GDSS, define the modules of group-decision as

cognitive, group organizing, decision-making by cooperation, feedback and adjust decision, conduce

consensus decision by negotiation, knowledge management and repository evolution, and explain the

process of every part.

1 INTRODUCTION

In most organizations, decisions are made by group

members who usually consider a set of attributes.

Group decision-making is considered to be a process

for deriving a single group preference from a

number of individual preferences regarding a set of

criteria and alternatives (L. Mikhailov, 2004). But it

is sometimes difficult to achieve a consensus among

group members. Therefore, a group

decision-supporting tool is needed to help group

members to reach a consensus for the group

decision-making under multiple attributes.

2 GDSS AND MULTI-AGENT

Group decision support system (GDSS) is an

interactive com

puter-based system and it combines

communication, computing, and decision support

technologies. It facilitates the solving of

unstructured or semi-structured problems by a group

of decision-makers (DeSanctis G, Gallupe RB,

1987), assists managerial decision-making by

presenting information and interpretations for

various alternatives and facilitates communication

among team members, regardless of the

geographical limitations and group decision

obstruction, so such system can help the

decision-makers to make more effective and

efficient decisions (Radermacher, F. J., 1994).

Group decision is a dynamic and continuous

rocess under conflict restraint. Problems in these

fields are usually semi-structured, and concern

complex, uncertain, incorrect, changing, and large

amounts of information. The critical and distinctive

feature of a group decision support system is to use

mathematical models, especial optimization models,

for decision-making and pose consensus decision.

As researchers began GDSS experimental studies,

a go

od decision generally cannot be defined by a

single criterion. It is not always the one with the

highest profit or the lowest risk. Generally it is some

combination of these two with other criteria such as

prestige, power, and ethical concerns. Equal

opportunity for participation appeared as a major

consideration in settings where groups were required

to reach a consensus (Watson et al., 1988;

Chidambaram et al., 1990; Easton et al., 1990;

Miranda and Bostrom, 1993).

The most charming advantage of GDSS is the

teraction and cooperation mechanism comparing

with traditional systems in which the individual

parts are designed independently with little

interaction, due to the lack of a unified

representation, simulation, and synthesis framework.

A number of typical characteristics of an intelligent

ent include: autonomy, proactiveness,

purposefulness, competence, reasoning capability

and interaction with environment or other Agents.

338

-Weiren S., -Daoping J., -Yonglin L. and -Jing C. (2004).

A GROUP DECISION SUPPORT SYSTEM - A description on models and modules in GDSS based on cooperative MAS.

In Proceedings of the First International Conference on Informatics in Control, Automation and Robotics, pages 338-341

DOI: 10.5220/0001135003380341

 SciTePress

An intelligent Agent is a software object capable of

communicating with other intelligent Agents, as

well as with humans, with a view to achieving a

given task. It monitors the world, anticipates the

consequences of its actions and the actions of other

Agents, and determines the action plan. An

intelligent Agent offers many advantages in the

situations that the GDSS cannot cover the entire

decision process. These features make intelligent

Agents an attractive tool for building active GDSS

(Franklin & Graesser, 1997; Luck, Griffiths, &

D’Inverno, 1997; Maes, 1995; Wooldridge &

Jennings, 1995).

There have been several attempts to combine

GDSS and Agent technology. A lot of researches

focus on combining GDSS and Agent technology to

improve decision quality.

Some researchers (Kuhlmann T, Lamping R,

Massow C, 1998) propose Agentified DSS software

to support decision-making. Bui et al. (Bui T, Lee J.,

1999) advocated an Agent based DSS framework to

facilitate decision-making in large-scale enterprises

and among scattered organization units. Paulson and

Kim at Stanford University conducted a study of

Agent-based project scheduling and control to

develop a software Agent-based system to support

decentralized decision-making. Some commercial

Agent-based GDSS have been presented, such as

Agnetis Business Systems in Australia, and Scientia

in British, which use distributed intelligence

technology for scheduling. Salo (Salo AA, 1995)

developed an interactive approach for the

aggregation of group members' preference

judgments in the context of an evolving value

representation. He suggests strict dominance (SD) or

weak dominance (WD) relations and in the case of

WD, the results are presented to the group members

and additional preferences are elicited for further.

Both GDSS and multi-Agent are distributed

systems and are good at deal with complex issue.

Normally, complex decision making tasks cannot be

done by a single Agent. Rather, they are typically

achieved through a coordinated effort of many

Agents with different sets of expertise and

assignment. We view GDSS based on multi-Agent

intelligent system as typically a network with

decision-making unites as nodes and communication

channels as links.

3 MODELS

As the feature of the system mentioned above, we

suggest the models of the GDSS based on

multi-Agent. We define the group decision models

based on Agent as:

Agent<S, D, A, G, C, P>

S: state of Agent, busy or idle. Idle Agent, not

the busy one, is chosen when facing with new issue.

It is useful in balance the system load, mentioned in

4.2.

D: domain of Agent. It is helpful in chosen the

right Agent to cooperation.

A: ability of Agent, input/out constraint, quality,

duration or cost. Only match ability of Agent with

goal of the issue can guarantee the veracity of the

decision.

G: goal of Agent. We can find the credit of

Agent by compare goal of Agent with the decision

outcome.

C: credit of Agent, expressed history record.

Agent with good credit can result in re-cooperation

by other Agents.

P: protocol of Agent accepted. Protocol of Agent

is the precondition of cooperation.

Issue<P, T, G, C, M >

P: priority of issue, important, urgency, or both.

The system pursues the important and urgent issue

in time.

T: deadline of the issue. The settlement fails if

the time is out.

G: goal of issue. The organized form of group

member determinates by goal of issue.

C: output constraint of issue. That is the

constraint on the group-decision outcome.

M: group members deal with the issue. It

expresses the group members that cooperate in the

issue decision-making.

4 MODULES OF THE GDSS

BASED ON AGENT

4.1 Cognitive Module

Cognitive behaviours play an important role

in-group decision process. Information and dataflow

with different forms exchange in several portions of

GDSS including input part, output part, feedback

part, even decision-environment and users. The

information and dataflow expressing change in

decision-environment or concession of the users

can’t identify by functional Agent. Real-time

responses take place when the change in

environmental is cognized. The group

decision-making will be interrupted or terminated

without cognitive behaviors.

A GROUP DECISION SUPPORT SYSTEM - A description on models and modules in GDSS based on cooperative MAS

339

4.2 Group organizing Module

The frames in GDSS we propose is not fixedly

organize Agent to execute. The system forms the

decision-group flexibly according to the two

principles: balance the system load and match the

capability of Agent with issue mission. On the one

hand, the Agent with different function distribute in

the GDSS, some Agents are overloaded while others

are idle all the time. Search the matched group

member from the idle Agents at first; turn to the

busy Agents if no Agent can be matched. All the

Agents in GDSS have the equal opportunity to

participate in the item. On the other hand, the

capability of Agent chosen must match with the

object of the issue, the perfect decision depend on

the correct match mechanism greatly.

4.3 Decision-making Module

Facing with the complexity of the issue, the

localization of computational capacity, no Agent can

fulfil the decision by itself. Due to the limitation of

information and resource availability, the

cooperation among group member is indispensable.

We view the procedure of model resolved and

decision come into being as cooperation among all

the Agents in the decision-group. Cooperative

systems are typically designed to perform complex

processes. In such systems, the effectiveness of a

GDSS to support them depends on the interaction

capacity and computational capacity of the

individual Agents.

4.4 Feedback Module

Traditional decision models in GDSS focus on the

methods of making group decisions, forming a

group decision through synthesizing group

members’ preferences while not paying much

attention to the problems of the environmental

change during the decision making period and

continual decision making under conflict restraint

(P.Lehner et al., 1997).

Feedback from the group points of view and to

the representations of the individual’s points of view

is required. Feedback means that a group member

changes his or her preferences so that they converge

to the other group members’ preferences, as

perceived by that member, or the

decision-environment changed (Pinson SD, Louca

JA, Moraitis P, 1997).

The group member changes the structure of his/

her evaluation system (e.g. changing the attention

paid to the criteria); and/or

The group member alters his/her own set of

criteria, adding or deleting some criteria (it is also

possible to create a criterion named group’s

opinion); and/or

The group member adjusts his/her assumptions

and predictions (subjective or not) according to

some extra information provided by the other

group’s elements; and/or

The group member uses another method or

process of aggregating the criteria that he or she

considers relevant; and/or

The decision-environment parameters changed

with the change in time, resource and so on.

Continuous decision is usually adjusted to reach a

better outcome through properly responding to and

managing the feedback.

4.5 Conduce consensus decision

Module

Conflict happen when system is short in source,

disagreement appears in target or alternatives

produced in group decision. A lot of research

focuses on the individual preference into group

preference. The compromised decision produces,

and so many group members is unsatisfied with the

result. The procedure of preference aggregation

violates the procedure of concede in group members

in nature.

Negotiation in GDSS takes place when the

requirements cannot be satisfied and an alternative

solution is to be provided. On the one hand, each

member in group would like to reach some

agreement rather than disagree and not reach any

agreement. But, on the other hand, each group

member would like to reach an agreement that is as

favourable to it as possible (Shaheen S. Fatima,

Michael Wooldridge, Nicholas R. Jennings, 2004)

Negotiation in our framework is a method of

Agent-oriented interactive problem solving. It is

clear that group decision support systems with the

negotiation support facility will be a key issue in the

next decade (M. Norita, 2000). The research on

negotiation focus on the four aspects: the negotiation

protocol; the negotiation strategies; the information

state of Agents; the negotiation equilibrium.

The newly emerging constraint Agent technology

provides a promising solution for such negotiation

Agents. Constraint Agent technology has emerged

as a promising research field. Constraint languages

are powerful tools to support the development of

constraint Agent systems. They provide necessary

facilities of representation, reasoning and

maintenance of constraint-based knowledge bases.

Although some difficult tasks need sophisticated

negotiation methods, interactive constraint

ICINCO 2004 - INTELLIGENT CONTROL SYSTEMS AND OPTIMIZATION

340

satisfaction is sufficient for most of GDSS tasks.

Sophisticated negotiation strategies can be

programmed with no great effort with flexible

constraint programming. For example, requirements

can be naturally represented as constraints and

objective functions of a set of variables. In addition

to the constraints about the user’s requirements, an

answer to the user may also have to satisfy the

constraints from the GDSS system, such as

constraints on the resources of the systems, and the

constraints about the relations among the Agents.

In one words, negotiation is necessary to balance

the satisfaction among different group members and

conduce the consensus decision to fulfil the

requirement of users.

4.6 Knowledge management Module

Knowledge used in decision-making is a valuable

asset of an organization, and accumulating such

knowledge is an important task (Kung-Jeng Wang,

Chen-Fu Chien, 2003). It is necessary to provide a

mechanism to store decision-related knowledge. In

this context, background decision-related knowledge

can be summarized as business rules and facts in

repository of every Agent to improve the quality of a

decision.

The decision-making process itself results in

improved understanding of the problem and the

process, and generates new knowledge. In other

words, the decision-making and knowledge creation

processes are interdependent. Because of such

interdependence, the research in the fields of group

decision support systems (GDSS) should integrate

the knowledge management systems (KMS) and

evolution in knowledge repository adequately.

5 CONCLUSION

The rationality to combine multi-Agent technology

and GDSS has been proved by the references

mentioned. We propose the models of Agent and

issue; suggest the function of the modules in GDSS.

The variables in the models are concise and simple.

The cognitive on the decision environment change;

procedure of group member organized; the

decision-making by cooperation; environmental

information by feedback; the procedure of

consensus decision conduce by negotiation and the

procedure of knowledge management can be

expressed by the models. Hence, the proposed

method is a promising and attractive alternative to

construct GDSS based on multi-Agent.

REFERENCES

Bui T, Lee J., 1999, An Agent-based framework for

building decision support systems, Decision Support

Systems, 25(3): 179–250.

DeSanctis G, Galluple RB, 1987, A foundation for the

study of group decision support systems, Management

Science, 1987 33(5): 589–609.

Franklin & Graesser, 1997; Luck, Griffiths, & D’Inverno,

1997; Maes, 1995; Wooldridge & Jennings, 1995.

Hans Tonino, Andr, Bos, Mathijs, de Weerdt, Cees

Witteveen, 2002, Plan coordination by revision in

collective Agent based systems, Artificial Intelligence,

142 (2002) 121-145.

Huaiqing Wang, Stephen Liao, Lejian Liao, 2002,

Modeling constraint-based negotiating Agents,

Decision Support Systems, 33 (2002) 201– 217.

Huaiqing Wang, Stephen Liao, Lejian Liao ， 2002,

Modeling constraint-based negotiating Agent ，

Decision Support Systems, 33 (2002) 201– 217.

Joao P. Costa, Paulo Melo, Pedro Godinho, Lu C. Dias,

2003, The AGAP system: A GDSS for project

analysis and evaluation, European Journal of

Operational Research, 145 (2003) 287–303.

Ki Jeong Chun, Hung Kook Park, 1998, Research

Examining the conflicting results of GDSS research,

Information & Management, 33 (1998) 313-325.

Kuhlmann T, Lamping R, Massow C, 1998, Intelligent

decision support, J Mater Process Techno, 76(1–3):1–

30.

Kung-Jeng Wang, Chen-Fu Chien, 2003, Designing an

Internet-based group decision support system,

Robotics and Computer Integrated Manufacturing, 19

(2003) 65–77.

L. Mikhailov, 2004, Group prioritization in the AHP by

fuzzy preference programming method, Computers &

Operations Research, 31 (2004) 293–301.

M. Norita, 2000, Folding arguments: a model for

representing conflicting views of a conflict, Group

Decision and Negotiation, 9 (2000) 63– 68.

Pinson SD, Louca JA, Moraitis P, 1997, a distributed

decision support system for strategic planning.

Decision Support Systems, 20 (1): 35–51.

P. Lehner et al., 1997, Cognitive biases and time stress in

team decision-making, IEEE Trans, 27（5）698–703,

September.

Radermacher, F. J., 1994, Decision support systems: scope

and potential. Decision Support Systems, 12, (1994)

257–265.

Salo AA, 1995, Interactive decision aiding for group

decision support, European Journal of Operational

Research, 84-134.

Shaheen S. Fatima, Michael Wooldridge, Nicholas R.

Jennings, 2004, An agenda-based framework for

multi-issue negotiation, Artificial Intelligence, 152

(2004) 1-45.

A GROUP DECISION SUPPORT SYSTEM - A description on models and modules in GDSS based on cooperative MAS

341