Constructing conceptual models for knowledge capture and delivery systems

Appropriate conceptualization of the nature of entities and relationships in a problem domain is a key prerequisite to the successful design of computerized decision aids for business, especially those developed for more than one idiosyncratic user. The need for a reliable conceptual model is particularly acute in the design of decision support systems that must function in problem‐solving situations with no existing theoretical framework or where theory and practice differ considerably.

This paper presents an interative procedure for developing a reliable conceptual model by testing the “fit”; of successive revisions of the model against a collection of “think‐aloud”; verbal problem‐solving protocols of people with experience in the domain. The model is revised each time until it is verifiably and consistently accurate. Such a procedure, it is argued, is more objective and reliable than intuition or traditional knowledge engineering and requires considerably less experimental data collection and analysis than more elaborate empirical model development procedures.

The feasibility of this procedure is illustrated by using it to construct a conceptual model for a computer‐based system that seeks to capture knowledge used during project planning and deliver it for use during project control.