A multi-agent systems approach to construction claims negotiation

Claims negotiation plays an important role in construction claims settlement and disputes resolution. However, claims negotiations are normally conducted inefficiently. Although many research projects have been undertaken on human behaviours in negotiation (e.g. negotiation planning, documentation and negotiation strategies) and computer-aided negotiation, there is not an effective approach to solving such problem. The development of multi-agent systems provides an innovative approach to facilitating claims negotiation, where intelligent agents can negotiate with each other for the real world parties that they represent. The significance of multi-agent systems lies in the fact that they match the fragmented nature of the construction industry. This thesis describes the work of developing a multi-agent system for construction claims negotiation (MASCOT). The objectives are to create an architecture for the agent system, and develop a negotiation mechanism for agent interaction. A conceptual MASCOT model is designed based on a thorough analysis of the nature, characteristics and problems of construction claims negotiation, multi-agent systems negotiation mechanisms, and negotiation theories. A modified Monotonic Concession protocol and the related negotiation strategies which are based on the integration of Zeuthen's risk evaluation model and Bayesian learning model were developed. A prototype was built using the ZEUS agent building toolkit and Java. The system was then assessedin terms of the quality of the negotiation mechanisma nd prototype using critical evaluation criteria and prototype evaluation. The result revealed that the MASCOT system could significantly enhance the efficiency of construction claims negotiation. Furthermore, it is recognised that multi-agent systems have a great potential to solve the fragmentation problem in other construction areas such as scheduling, concurrent engineering; and collaborative design, particularly when the project team members are geographically distributed. This research not only contributes to the improvement of construction claims negotiation, bdt also provides an effective approach for the development of multi-agent 'system negotiation mechanism.