A strategy for evaluating a fuzzy case-based construction procurement selection system

Abstract

System Verification and Validation (V&V) is an essential element in the development and implementation of any computer-based decision tools. The unique concepts of Case-Based Reasoning (CBR), such as the use of mega-knowledge and nearest matching have generated extra challenges to system developers to ensure that the system is built right and the right system is built. However, little attention has been attributed to verifying and validating a CBR system. Recently, a fuzzy CBR prototype known as CaPS has been developed for the selection of appropriate construction procurement systems. To ensure that the procurement system is acceptable to the procurement experts in the construction industry, a series of tests have been conducted with domain experts using real cases (stored in the case base) and projects (as scenarios for retrieval and comparison). This paper reports on the findings of the V&V that have been performed on CaPS. Techniques available for verifying and validating a CBR system are first discussed. The V&V procedures applied to the prototype system are subsequently outlined. The results confirm that the cases stored in CaPS are correct, consistent, and irredundant. More importantly, the solutions generated by CaPS are accurate and innovative, and these are necessary for today's construction projects. © 2005 Elsevier Ltd. All rights reserved.