A methodology and operationalization for acquiring knowledge from multiple experts
Sathasivam Mathiyalakan, James R. Marsden · 1995
When there are multiple experts in an organization, or when a system development project requires the use of multiple experts, there is a need to determine the better group size for knowledge acquisition, as the group size may affect the quality of knowledge elicited. The primary theme of this research effort is to investigate issues relating to the capabilities of groups of different sizes in developing expert systems dealing with decision making tasks. Our research methodology is organized into two stages. In the first stage subjects act alone and our purpose is to select candidates for subsequent group sessions. In the second or group stage, we study the relationships among group size, decision rules, availability of what-if capability, change in group membership, performance, decision time, productivity and satisfaction. A key feature of our study is the use of market conditions, that is the use of performance based rewards along with system usage costs. To achieve our research objectives we develop programs written in the C language, one in stand-alone mode for the individual setting and one in network mode for the group setting. Prototyping, a system development technique is used to design the experimental screen, software and to set experimental parameters. A senior officer from a leading bank, faculty members and volunteer students provided significant help during the system development phase. Our subject pool consists of student subjects and bank personnel. There are two major findings in this study, both of which can serve as recommendations for practitioners. First, we note that using a market approach subjects are able to provide a realistic assessment on the nature and value of the support provided. Second, we find no dominant group size. We find that each group size that we examine has its benefits and drawbacks. Finally, we identify issues that a Knowledge Engineer needs to consider in selecting a group size for use in rule set development for building an expert system that provides decision making support.