Searching for better negotiation agreement based on genetic algorithm

Abstract

Negotiation is commonly required in the procurement of construction materials to reach the final contractual agreement. In current practice, contractors negotiate with suppliers according to negotiators' experiences instead of extensive exploration of negotiable options and negotiators' preferences. Consequently, negotiators often reach suboptimal agreements, and leave money on the table. This research intends to help negotiators explore negotiable options by developing a computer system, named C-Negotiator, using the genetic algorithm. This article also describes experiments conducted to determine how much money was left on the table on typical realistic construction procurements. The result shows that C-Negotiator's negotiation improved the joint payoff of the contractor and supplier from 1.5% to 9.8% compared with conventional human negotiation. The improvement may increase for more complex negotiation problems with more options and complicated preferences or for inexperienced negotiators.