Multi-Agent Architecture for Knowledge-Driven Decision Support - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 44

such as the Internet, is a challenging task; yet, realizing this task can have significant benefits for orga-
nizations embracing such collaborations. An application of the IKMDSA for Collaborative Commerce
to enable collaborative work in B2B e-Marketplaces would have significant benefits in developing
information partnerships by creating the foundation for knowledge representation and exchange by
intelligent agents that support collaborative work between business partners.

conclusion, l iMit ations and f uture direction for r esearch

In this chapter we have presented a methodology to represent modular, rule-based knowledge using the
extensible markup language (XML) and the document type definition (DTD) standards from the World
Wide Web Consortium (W3C). Using this methodology, we have shown how such an approach can be used
to create problem-specific knowledge modules that can easily be distributed over the Internet to support
distributed IDSS design. Such an approach will facilitate intelligent decision support by providing the
required knowledge representation and the decision analytical support. We had presented the conceptual
architecture of such a distributed IDSS, and have provided details of the components of the architecture,
including the agents involved and their interactions, the details of the knowledge representation and
implementation of knowledge exchange through a distributed interface. We also provided indication of
how such architecture might be used to support the user and how it might assume the role of an expert
and provide explanations to the user, while retaining the benefits of an active DSS through extensible
knowledge generation by incorporating machine learning algorithms. The example used in this chapter
is simple, intuitive and elegantly achieves its purpose of illustrating the use of the architecture while
minimizing complications inherent to a more complex problem domain. We continue to do research
on elaborating this architecture for a variety of problems that lend themselves to rule-based, inductive
decision making with a need for user interactions and which benefit from greater understanding of the
problem domain by the user.
The limitations of this research derive from the use of decision trees and inductive learning al-
gorithms and techniques. The limitations inherent to decision trees and such techniques are also the
limitation of this architecture. Therefore, further research needs to be conducted to understand how this
architecture can be expanded to incorporate other types of learning and rule induction or rule creation
to be shared and used by software agents. Despite this limitation, this chapter contributes significantly
to the advancement of our understanding of how emerging technologies can be incorporated into intel-
ligent agent-based architecture to enhance the value of such systems in distributed intelligent DSS that
incorporates knowledge.

r eferences

Adriaans, P., & Zantinge, D. (1996). Data mining. Harlow, UK: Addison-Wesley.

Apte, C., & Weiss, S. (1997). Data mining with decision trees and decision rules. Future Generation
Computer Systems, (13), 197-210.