WEB SERVICES-BASED INTEGRATION OF HETEROGENEOUS DATABASES USING A MULTI-AGENT SYSTEM Hazem T. El-Khatib, College of Information Technology, United Arab Emirates University, Al Ain, P.O.Box 17555, United Arab Emirates. M. Howard Williams, David H. Marwick and Lachlan M. Mackinnon School of Math and Computer Sciences, Heriot-Watt University, Edinburgh EH14 4AS, UK ABSTRACT An important application area in which software agents have an essential role to play is that of providing user access to a collection of heterogeneous distributed databases. This paper describes a general approach, which uses a multi-agents system based on Web services and common object request broker architecture (CORBA) to retrieve information from a set of heterogeneous distributed databases and consolidate it, taking account of the syntactic and semantic differences between data sources. The system is dynamic in that the data sources accessed by the system can change with time. The ability of software agents to negotiate and co-operate with each other make them ideal candidates for this kind of system. Web services can move software agents to the Internet. The role of CORBA in such a system has been investigated. Initially CORBA was used to enable software agents to access each others' services - although this limits the flexibility of the system. A better solution is based on the use of Web services while CORBA is used as the technology to implement the business logic for the software agent, and will be exposed to the Internet using Web services. KEYWORDS Software Agent, Web Services, Heterogeneous Databases. 1. INTRODUCTION Distributed database development has been an important field ever since computing and data storage moved from monolithic architectures and encapsulated jobs on centralized mainframe computers to client/server architectures on peer-networked workstations and subsequently to n-tier architectures. The distribution of databases and services across multiple workstations evolved as a way to scale and organize the data centre using a variety of approaches to communications and low-level facilities for integration with arbitrary applications. The information systems strategies of enterprises moved from database centric to the perspective of information flow centric leading to a new generation of business services. However, this approach requires those business logics scattered over the Web to be made accessible in the form of services available throughout the Internet. This paper describes the problem of integrating information from heterogeneous distributed databases in response to a user query. This problem is complex and it has attracted a considerable amount of attention over the years and has been investigated by a number of researchers. Various papers have been published describing a variety of different approaches, which have been developed to handle this problem, although aspects of the problem remain. Surveys and comparisons of systems developed to handle this problem can be found in [Thomas, Thompson, et al.] [Litwin, Mark, et al]. This paper describes an on-going system that integrates and retrieves information from a collection of databases using Web services that are encapsulated within software agents. It distributes the task of resolving heterogeneity between autonomous and co-operating agents. Moving knowledge from a centralised II - 253