ANN-based mark-up estimation system with self-explanatory capacities

Artificial neural networks (ANNs) have been applied to support construction mark-up estimation. The major drawback of this application, however, is that an ANN system is unable to explain why and how a particular recommendation is made. This significantly affects the user-acceptance of the system and its results. The research presented in this paper investigates the use of the KT-1 method for automatically extracting rules from a trained neural network. The KT-1 method is implemented and tested on collected bidding data, and the results from the investigation indicate the usefulness of the KT-1 method. Discussions on the difficulties of generating automated explanations are also presented.