Managing expert systems projects: factors critical for successful implementation

• Authors: Dave Dibble;Robert P Bostrom
• Venue: SIGCPR '87 Proceedings of the conference on The 1987 ACM SIGBDP-SIGCPR Conference
• Year: 1987

Over the last decade, the Artificial Intelligence research community has successfully created a number of “expert” systems capable of solving problems in limited domains. Noting the demonstrated feasibility of such an approach to problem solving, a number of firms have moved toward incorporating this technology into their portfolio of information system tools. The scarcity and geographic limitations of human expertise, coupled with the tirelessness of machines, are the motivations suggested as reasons organizations investigate the technology. In fact, a number of real world applications have already been developed and placed into daily operation.Successful management of the systems development process (from design through implementation and evaluation) is critical to the success of an Information Systems (IS) organization. The evolution of new information technologies tend to complicate this process. Expert Systems (ES) is a new technology with development characteristics of its own. This paper addresses the development issues faced by the IS manager who is considering using this technology in a business context. It is understood that ES technology itself will continue to grow and change. But if it is to become a viable IS technology over the long run, it will have to achieve some degree of manageability.Because of the newness of the technology, little research has been conducted on the management of ES projects. Instead most of the effort has been focused on various attributes (uncertainty, different reasoning techniques, structuring of the knowledge, etc.) of ill-structured problems, and how ESs could be constructed to deal with these attributes. Still, despite the lack of rigorous research on the management of such projects per se, the literature does suggest a number of variables crucial to successful implementation. This paper discusses a framework based on existing systems development models, within which the management of ES projects can be considered. This framework will serve to organize the issues, relative to the development of ES, that have been reported to date. It will also suggest other important issues not yet reported in the literature.The framework is used to categorize existing lines of research and to discuss how they could be appropriately extended. Since the interest is the use of ESs in a business context, the focus will be on those research issues best addressed by the IS community.