Percentage of procedural knowledge acquired as a function of the number of experts from whom knowledge is acquired for diagnosis, debugging, and interpretation tasks

It is indicated that the acquisition of procedural knowledge can significantly be improved, depending on the tasks (interpretation, debugging, or diagnosis) for which the knowledge is elicited; the method of knowledge elicitation (protocol, interview, induction, or repertory grid), and the number of experts (from one to six).

Abstract

Knowledge elidtation is an expensive and vital part in the construction of knowledge‐based systems. This paper indicates that the acquisition of procedural knowledge can significantly be improved, depending on the tasks (interpretation, debugging, or diagnosis) for which the knowledge is elicited; the method of knowledge elicitation (protocol, interview, induction, or repertory grid), and the number of experts (from one to six) from whom knowledge is elicited. On average, for all conditions the percentage of acquired procedural knowledge increased by two‐fold from using one to six experts. The optimal cost‐benefit utility may be achieved by using three experts for knowledge elicitation.