A multiagent approach for diagnostic expert systems via the internet Khaled Shaalan a, * , Mona El-Badry b , Ahmed Rafea c a Department of Computer Science, Faculty of Computers and Information, Cairo University, 5 Tharwat Street, Orman, Giza 12613, Egypt b Central Laboratory for Agricultural Expert Systems (CLAES), P.O. Box 100, Dokki, Giza, Egypt c Department of Computer Science, American University, 113, Sharia Kasr El-Aini, P.O. Box 2511, 11511 Cairo, Egypt Abstract In recent years there has been considerable interest in the possibility of building complex problem solving systems as groups of co- operating experts. This has led us to develop a multiagent expert systems capable to run on servers that can support a large group of users (clients) who communicate with the system over the network. The system provides an architecture to coordinate the behavior of several specific agent types. Two types of agents are involved. One type works on the server computer and the other type works on the client computers. The society of agents in our system consists of expert systems agents (diagnosis agents, and a treatment agent) working on the server side, each of which contains an autonomous knowledge-based system. Typically, agents will have expertise in distinct but related domains. The whole system is capable of solving problems, which require the cumulative expertise of the agent community. Besides to the user interface agent who employs an intelligent data collector, so-called communication model in KADS, working on the client sides. We took the advantage of a successful pre-existing expert systems—developed at CLAES (Central Laboratory for Agricultural Expert Systems, Egypt)—for constructing an architecture of a community of cooperating agents. This paper describes our experience with decomposing the diagnosis expert systems into a multi-agent system. Experiments on a set of test cases from real agricultural expert systems were preformed. The expert systems agents are implemented in Knowledge Representation Object Language (KROL) and JAVA languages using KADS knowledge engineering methodology on the WWW platform. q 2003 Elsevier Ltd. All rights reserved. Keywords: Expert systems; Agents; WWW 1. Introduction As computer hardware and software become increas- ingly powerful, so do applications, which used to be considered beyond the scope of automation. To cope with increased demands, software systems are becoming correspondingly larger and more complex (Genesereth & Ketchpel, 1994; Jennings & Varga, 1993; Jennings & Wooldridge, 1995). In recent years, there has been considerable interest in the possibility of building complex problem solving systems as groups of co-operating experts. Distributed Artificial Intelligence (DAI) is the study of how such systems might be built. Historically, empirical research in DAI has focused on three main areas: blackboard architecture (Engelmore & Morgan, 1988; Nwana, 1996), systems based on negotiation (Smith, 1980), and multi-agent planning systems (Durfee and Lesser, 1988). More recently, co-operating expert systems have emerged as a research area of some importance (Nwana & Ndumu, 1999). Agent-based computing represents an exciting new synthesis both for Artificial Intelligence (AI) and, more generally, Computer Science. It has the potential to significantly improve the theory and the practice of modeling, designing, and implementing computer systems (Jennings, 2000). Since the 1980s, software agents and multi- agent systems have grown into what is now one of the most active areas of research and development activity in computing generally (Jennings & Wooldridge, 1998). There are many reasons for the current intensity of interest, but certainly one of the most important is that the concept of an agent as an autonomous system, capable of interacting with other agents in order to satisfy its design objectives, is a natural one for software designers (Wooldridge & Ciancar- ini, 2001). Just as we can understand many systems as being composed of essentially passive objects, which have state, 0957-4174/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2003.12.018 Expert Systems with Applications 27 (2004) 1–10 www.elsevier.com/locate/eswa * Corresponding author. Tel.: þ 2-012-2223377; fax: þ 2-02-761-7628. E-mail addresses: shaalan@mail.claes.sci.eg (K. Shaalan); mona@ mail.claes.sci.eg (M. El-Badry); rafea@aucegypt.edu (A. Rafea).