Implementation Of A Negotiation Model Based On The Principle Of Leverage

In real life, a negotiation is defined as a conversation between people trying to find a common understanding on issue. The issue of negotiation can be anything such as price, peace, brides, etc. Most of the time, negotiation as a conversation involves two sides that agree to observe basic communication skills for the negotiation to be possible. The most common negotiation process is found in price bargaining such as auction where buyer and seller continuously exchange offers and counter-offers until the final price of a good is found. Researches in multi-agent technology exposes different negotiation methods of negotiation related to price bargaining. In most cases, buyer and seller are modeled as agent with strategies and tactics that allow them to cooperatively or uncooperatively move towards a joint solution that either agent fully accepts. The final price of a good lies within a negotiation interval, which is limited by reservation prices specified by buyer and seller at the beginning of negotiation. The price of the good is accepted by either agent because it satisfies its payoff function at the end of negotiation. Indeed, when acting upon environment, the aim of agents such as buyer and seller is to satisfy its welfare within the environment.Though previous researches provide strategies and tactics that allow buyer and seller agents to decide a final price of a good within the interval of negotiation, nothing prevents this final price to be one of the reservation prices. We depart from this fact and expose a new way of allowing buyer and seller agents to fully maximize their payoff functions throughout and at the end of negotiation. With the game theoretic, we define a new game based on the principle of leverage whereby buyer and seller continuously track the final point of a negotiation. Based on mechanism design, we develop game rules and pre-condition to establish an enforcement principle at the fulcrum point, which corresponds to the final turn of negotiation. From these rules and pre-conditions, we define the Law of Lever function or LL. The move towards the final point of negotiation is assured by Time Dependent Tactics (TDT), which are functions that has got a special property of being maximized to their reservation prices at the end of negotiation. In addition, based on the information set of the buyer and seller i.e. possible outcome of the game at each turn of negotiation, we establish the payoff function of each agent involved in negotiation. Having defined TDT, LL and payoff functions, we design tactics undertaken by each agent upon receiving a response from the other agent. Furthermore, by combining Time Dependent Tactics, LL functions, and payoff functions we allow either of the two agents to accept or refuse the final outcome, which results in agreement or disagreement. Negotiation phases are designed based on a modified version of the Contract Net Protocol.The negotiation model is successfully verified through its implementation on JADE platform and having Model-View-Controller as architecture. Since negotiation messages are not instantaneously exchanged between agents, time losses are observed during negotiation. Thus the overall negotiation duration is slightly increased to compass these losses. Finally fairness of the negotiation model is proposed as the solution to agents who expose more Boulware or Conceder behaviors.