Premium
Developing an expert system on project evaluation: Pt II: Structuring the system
Author(s) -
Wilkinson A.
Publication year - 1991
Publication title -
randd management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.253
H-Index - 102
eISSN - 1467-9310
pISSN - 0033-6807
DOI - 10.1111/j.1467-9310.1991.tb00758.x
Subject(s) - structuring , context (archaeology) , process (computing) , computer science , work (physics) , work breakdown structure , quality (philosophy) , expert system , focus (optics) , process management , management science , engineering management , knowledge management , project management , engineering , project planning , systems engineering , project charter , artificial intelligence , business , programming language , mechanical engineering , paleontology , philosophy , physics , finance , epistemology , optics , biology
The work here described is the early stages of a programme to develop a series of Expert Systems (ES's) in the management of R&D. In particular, the development of a prototype ES using a package program Crystal Ver. 2.1 is the focus together with the problems encountered. The literature had suggested that the use of such packages facilitated the process for those exploring ES's. What was not clear from the literature or the program manuals was the limited range of problems which could be handled. The programs have a rigid structure such that unless the logic of the problem solving process matches that of the program, severe difficulties will ensue. This paper outlines the inherent structure of the programs and indicates the kind of problems which would fit. Specific problems which were encountered when trying to stretch the use of the program are explained in detail. It goes on to explain how these problems were got round using Lotus 1–2–3 which proved far more flexible in practice. This kind of understanding is essential if ES's are to become every‐day tools of R&D management. The objective in developing the Expert System (ES) was for it to be advisory to those preparing project proposals for possible submission to higher management within the context of an R&D laboratory. Thus the intention was to capture the expertise of senior managers, make it available to more junior ones, and thereby improve the quality of their work. This concept proved to be naive in that the expertise proved to be so diffuse that a wholly new rationale to structure the expertise had to be developed. This process, the subsequent development into a prototype ES, and its evaluation, have already been published (Wilkinson, 1991). The literature on ES's had suggested that the use of available package programs was straight forward and that the real problem was in establishing the expertise to be used and in structuring it to fit the program. This too proved to be an over‐simplification. When this work started, the exact nature of the package programs had not been spelt out adequately. The purpose of this paper, Part II, is to describe their inherent structure and the limitations which this imposes on the kind of problems for which they can be used. The insights gained in this work into the evaluation of projects and the inadequacies of previous evaluation techniques will form the contents of Part III of this series of papers.