基于杠杆原理的协商模型的应用研究

论文摘要

在现实生活中,协商指协商双方通过交流、沟通企图在一特定的问题上达成共识。任何事物都可以进行协商,如价格、和平、新娘等。多数情况下,协商被认为是谈判双方在遵守基本谈判规则下进行的沟通交流。最常见的协商常在议价过程中表现出来,比如拍卖,买卖双发会不断的进行价格协商直到一个合理的价格出现。目前,关于多Agent技术的研究所提出的不同协商方法大都与议价有关。在大多数情况下,卖家和买家被表示为有策略与战略的Agent,并且允许他们合作地或不合作地达成任何一个Agent都完全接受的联合解决方案。最终这个合理的价格应处在协商间隔内,该协商间隔是通过买方和卖方在谈判开始时的预定的价格进行限制的。然而,这个合理的价格能够被任何一个Agent接受是因为该价格满足了在谈判结束时的支付函数。事实上,当环境的发生变化,买方Agent与卖方Agent的目的就是在所处的环境中满足其相应的利益。虽然先前相关研究提供的Agent策略与战略是允许买方Agent与卖方Agent在协商间隔中决定最终的最优价格,但是这样并不能保证最终价格是最初的一种预定价格。我们从这一事实出发,提出了一种新的方法,允许买卖双发Agent在协商结束后最大化其相应支付函数。利用博弈论,我们定义了基于杠杆原理的游戏规则,即买方Agent和卖方Agent不断的跟踪协商的最终点。基于该原理的设计,在杠杆的支点通过制定游戏规则和先决条件来建立游戏执行原则。根据这些规则和先决条件,定义了杠杆函数准则(LL)。同时通过使用时间依赖策略(TDT)确定走向协商的最终点,TDT是协商结束后的预定价格被最大化的特殊属性函数。另外,根据游戏可能的结果,即买方和卖方在每个回合谈判的信息,建立了协商过程中相关的所有Agent的支付函数。论文首先定义了TDT、LL和支付函数,然后,根据每个Agent从其他Agent的接收到的响应来设计Agent策略,最后,通过使用时间依赖策略,LL函数和支付函数,允许每两个Agent接收或拒绝最终结果,即是否达成双方的协议。该协商阶段是在基于合同网络协议的修改版本上进行的。因在协商过程中,各Agent之间没有立即进行协商信息的交换,导致可执行时间的丢失。因此使得系统的全局谈判时间略有增加。本文提出的公平协商模型解决了拥有更多Boulware或Conceder行为的Agent所出现的这一问题。通过在JADE平台上,使用MVC架构进行实验,成功的验证该协商模型的可行性。

论文目录

• 摘要
• Abstract
• Dedication
• Acknowledgements
• Table of contents
• List of abbreviation
• List of tables
• List of figures
• Chapter 1:Introduction
• 1.1 Inroduction
• 1.1.1 Research motivation and background
• 1.1.2 Research hypothesis and problem formulation
• 1.1.3 Research objectives
• 1.1.4 Research contribution
• 1.1.5 Organization of the thesis
• 1.2 Introduction to negotiation and multi-agent negotiation
• 1.2.1 Why do people negotiate?
• 1.2.2 Need of automation in negotiation
• 1.2.3 Multi-agent negotiation
• 1.2.4 What is an agent?
• 1.2.5 Agent in negotiation processes
• 1.3 Conclusion of chapter one
• Chapter 2:Related Theory and Technologies
• 2.1 Game theory
• 2.1.1 What is a game?
• 2.1.3 Multi-agent and strategies
• 2.1.4 Utility and preferences in a game
• 2.1.5 Strategic game
• 2.1.6 Strategy dominance and game equilibrium
• 2.2 Auction
• 2.2.1 What is auction?
• 2.2.2 Different types of auction
• 2.2.3 Auction as structured negotiations
• 2.2.4 Mechanism design:can we design any type of auction?
• 2.2.5 Cooperative distributed problem solving
• 2.3 Contract Net Protocol
• 2.3.1 Actors roles in Contract Net Protocol
• 2.3.2 Basic concepts of the Contract Net Protocol
• 2.3.3 Assessment of the Contract Net Protocol
• 2.3.4 Principle of leverage
• 2.4 Related researches to our research
• 2.4.1 Previous research on multi-agent negotiation
• 2.4.2 Price bargaining in multiagent negotiation:a focus on researches that inspired our research topic
• 2.5 Conclusion of chapter two
• Chapter 3:Negotiation Model based on The Principle of Leverage
• 3.1 Enforcement in the law of levers
• 3.1.1 Informal description of a game based on the law of levers
• 3.1.2 The existence of an enforcement and its necessity
• 3.2 Multiagent negotiation model using the principle of leverage
• 3.2.1 Negotiation model
• 3.2.2 Negotiation process
• 3.2.3 Principle of leverage equation
• th step and the total number of unities'>3.2.4 Agent's number of unities at i^th step and the total number of unities
• i+1 or y_i+1'>3.2.5 Time Dependent Tactics: Generation of x_i+1 or y_i+1
• 3.2.6 Formal description of the regotiation model from a strategic game perspective
• i,t_i）and y（_i,t_i） moves towards the final negotiation round （i,t_i）'>3.2.7 Example showing how x（_i,t_i）and y（_i,t_i） moves towards the final negotiation round （i,t_i）
• 3.3 Conclusion Of chapter three
• Chapter4: Design and Analysis of the Negotiation Model
• 4.1 Description of OAS
• 4.2 Used platform: J2EE platform overview
• 4.2.1 Used Architecture:Model-View-Controller（MVC）architecture
• 4.3 Technologies
• 4.3.1 Java and JSP
• 4.3.2 JADE
• 4.3.3 ORACLE RDBMS
• 4.4 Analysis and Design Methods
• 4.4.1 Object Oriented Analysis and Design（OOA & D）
• 4.4.2 Agent Oriented Analysis and Design
• 4.5 Scope in design,analysis and implementation
• 4.6 Design and analysis of non-agent parts of OAS
• 4.6.1 Use case diagrams
• 4.6.2 OAS requirements
• 4.6.3 Subsystems of OAS
• 4.6.4 OAS Object Model
• 4.6.5 OAS data model
• 4.7 Developing agent part of OAS
• 4.7.1 System specification
• 4.7.2 Architectural design
• 4.7.3 Detailed design
• 4.8 Chapter four conclusion
• Chapter5: Implementation of the negotiation model
• 5.1 Initial negotiation model testing
• 5.2 Presentation of results from initial model testing
• 5.2.1 Agreement case
• 5.2.2 Disagreement case
• 5.3 Result discussion
• 5.4 Chapter five conclusion
• Chapter 6: Conclusion and future works
• References
• Research and publications