In this paper, we propose a Multi-agent model for web-based collaborative decision support system in which a facilitator and group decision makers are supported by agents. The integrated agents into web-based collaborative decision support system constitute a collection of autonomous collaborative problem solving intelligent agents, goal-directed, proactive and self-starting behaviour; interact with other agents and humans in order to solve problems. Specifically, agents were used to collect information and generate alternatives that would allow the user to focus on solutions that were found to be significant. The decision making process, applied to the boilers defects in an oil plant, relies on a cycle that includes recognition of the causes of a defect (diagnosis), plan actions to solve the incidences and, execution of the selected actions.
As organizations seek to adapt in a world of rapid change, decision making becomes increasingly dynamic and complex. Collaborative decision support systems provide a means by which a larger number of organizational stakeholders can efficiently and effectively participate in the decision making process. A greater number of organizational members participating in the decision making process logically leads to a better decision. The resulting decision should benefit by the richness of knowledge provided by the greater representation of organizational members. A success factor critical to this involvement is the successful organization of massive amounts of information generated by such a group.
On the other hand, the Distributed Artificial Intelligence (DAI), which is commonly
implemented in the form of intelligent agents, offers considerable potential for the
development of information systems and in particular Decision Support Systems (DSS).
Widely range applications domains, in which agent solution is suggested, are being applied
or investigated [
This article takes a multi-agent view of the web-based collaborative decision making process and examines the potential integration of agent technology into a distributed group decision support systems. It considers group participants as multiple agents concerned with the quality of the collaborative decision. We define a facilitator agent as that agent responsible for the overall decision making process. This includes managing the complex negotiation processes that are required among those participants collaborating on decision making.
We take first a literature survey of some related work in section 2 and 3. Then we propose a multi-agent architecture for web-based collaborative decision support systems in section 4. We also present some implementations issues in section 5. Finally, we conclude with future research direction in section 6. 2
Decision aid and decision making have greatly changed with the emergence of information and communication technology (ICT). Decision makers are now far less statically located; on the contrary they play the role in a distributed way. This fundamental methodological change creates a new set of requirements: web-based collaborative decisions are necessarily based on incomplete data. “web-based collaborative decision” means that several entities (humans and machines) cooperate to reach an acceptable decision, and that these entities are distributed and possibly mobile along networks. Distributed decision making must be possible at any moment. It might be necessary to interrupt a decision process and to provide another, more viable decision.
Collaborative or Group Decision Support Systems (GDSS), which are closely related to
DSS, facilitate the solution of unstructured and semi-structured problems by a group of
decision makers working togethe
Research that studied group decision support systems in the existing literature used mainly face-to-face facilitated collaborative decision support systems. Some of its results may not apply to distributed teams that, it is difficult for distributed teams to arrange face-toface meetings or to meet at the same time virtually. 3
In recent years, there has been considerable growth of interest in the design of a distributed, intelligent society of agents capable of dealing with complex problems and vast amounts of information collaboratively. Various researches have been conducted into applying intelligent agent-based technology toward real-world problems. Furthermore, there has been a rapid growth in developing and deploying intelligent agent-based systems to deal with real-world problems by taking advantage of the intelligent, autonomous, and active nature of this technology. The main benefits of an agent-based approach come from its flexibility, adaptability, and decentralization.
The definition of multi-agent systems (MAS) is well known and accepted as a loosely coupled network of agents that work together to find answers to problems that are beyond the individual capabilities or knowledge of each agent and there is no global control system.
An agent’s architecture is a particular design or methodology for constructing an agent.
Wooldridge and Jennings refer to an agent’s architecture as a software engineering mod
MAS are software systems composed of several autonomous software agents running in a distributed environment. Beside the local goals of each agent, global objectives are established committing all or some group of agents to their completion. Some advantages of this approach are: 1) it is a natural way for controlling the complexity of large and highly distributed systems; 2) it allows the construction of scalable systems since the addition of more agents become an easy task; 3) MAS are potentially more robust and fault-tolerant than centralised systems.
As is typical with an emerging technology, there has been much experimentation with the
use of agents in DSS, but to date, there has been little discussion of a framework or
methodological approach for using agents in DSS, and while DSS researchers are discussing
agents as a means for integrating various capabilities in DSS and for coordinating the
effective use of information [
We started our framework with the following fundamentals:
1. The first fundamental, in keeping with [
In [
In our proposed framework [
Agents were integrated into the DSS for the purpose of automating more tasks for the
user, enabling more indirect management, and requiring less direct manipulation of the
collaborative decision support system. Specifically, agents were used to collect information
outside of the organisation and to generate decision-making alternatives that would allow the
user to focus on solutions that were found to be significant. A set of agents is integrated to
the system and placed in the collaborative decision support system components, according to
our framework [
The goal of Distributed Group Decision making is to create a group of coarse-grained cooperating agents that act together to come to a collective decision. Participants in a collaborative decision making meeting are considered as a set of agents involved in creating a collective decision. These participant agents are involved with the content knowledge of the particular group problem at hand. The responsibility of managing any decision making process is typically put upon a supervisory agent. We call this agent the facilitator. We view the participants as multiple agents responsible for creating the content of the decision, and the facilitator as an outside agent responsible for managing the decision process that the participant agents use to come to common decision
For each participant (decision’s maker), the following agents are defined: • DA (Decision-maker Assistant): it’s the interface between the participant and the system. During idea (solution) generation stage, a decision maker can use its proper DSS (Decision Support System) through the DA.
• CA (Collaborator Assistant): The role of this agent is devoted exclusively to the collaboration of the decision maker in the process of decision making support. The only interaction it manages is with CRA of the facilitator and does not communicate directly with agents of other decision makers.
For the facilitator side, the following agents are defined: • FA (Facilitator Assistant): it manages the interface between the system and the facilitator. It provides a private workspace for the facilitator and a public space for the group. It also allows the facilitator to communicate at any time with group members outside the decision making process, helps to establish communications with other system users through their assistants (DA).
• CRA (CooRdinator Agent for the decision making process): It’s the central agent of the decision making process. It is supervised by the facilitator via the FA. Its role is to ensure the rules checking and application during the various phases of the decision making process. FA starts the decision making session. The CRA takes in charge the following tasks of this activity. It guides the group through the activity phases. • MA (Mediator Agent): is requested by the CRA during the alternatives organisation phase. Its role is to refine the alternatives (deletes or merges synonymous, redundant or inconsistent alternatives) and to classify the alternatives as well.
A prototype of the multi-agent architecture for distributed group decision support system is being implemented in order to generate results that can be analyzed and validate our work. To this end, we have used the FIPA compliant JADE platform to implement our system. Some implementation details are given in the next section.
As depicted in figure 4, a decision group composed of a facilitator and four decision makers collaborate and interact to solve a problem; the decision maker number three doesn’t appear on the figure as it’s disconnected and does not participate to the decision making session. A partial result of the interactions between agents (JADE’s sniffer screen) is given Figure 4. 6
In this paper we presented a web-based collaborative decision support system based on a multi-agent architecture. We have integrated agents into a cooperative intelligent decision system for the purpose of automating more tasks for the decision maker, enabling more indirect management, and requiring less direct manipulation of the DSS. In particular, agents were used to collect information and generate alternatives that would allow the user to focus on solutions found to be significant. Agents are normally used to observe the current situation and knowledge base, and then make a decision on an action consistent with the domain they are in, an finally perform that action on the environment.
The use and the integration of software agents in the decision support systems provide an automated, cost-effective means for making decisions. The agents in the system autonomously plan and pursue their actions and sub-goals to cooperate, coordinate, and negotiate with others, and to respond flexibly and intelligently to dynamic and unpredictable situations.