多Agent自动协商中机器学习的应用研究

目前将机器学习理论应用到多Agent自动协商系统中已成为电子商务领域的最新研究课题。本文即是利用贝叶斯法则来更新协商中的环境信息（即信念），利用强化学习中的Q学习算法生成协商中的提议，建立了一个具有学习机制的多Agent自动协商模型。并且封传统Q学习算法追行了扩充，设计了基于Agent的当前信念和最近探索盈余的动态Q学习算法。实验验证了算法的收敛性。     

多Agent自动协商的公平性研究

为了能够快速、高效地进行Agent协商,构建一个优化的多Agent协商模型。在这个模型的基础上,提出了一个基于协商各方公平性的协商学习算法。算法采用基于满意度的思想评估协商对手的提议,根据对方Agent协商历史及本次协商交互信息,通过在线学习机制预测对方Agent协商策略,动态得出协商妥协度并向对方提出还价提议。最后,通过买卖协商仿真实验验证了该算法的收敛性,表明基于该算法的模型工作的高效性、公平性。     

一个信任和声誉模型及其应用

购物Agent在基于Web的电子市场中携带着用户的需求选择合适的销售者,在信息不完全的情况下,购买者需要利用信任和声誉在销售者中选择自己信任并能给自己带来较高效用的销售者进行协商。论文提出了一个信任和声誉度量模型,并为购物Agent提出了一个建立信任和声誉模型的算法,购物Agent以协商历史为基础利用这个算法来选择适合自己的销售者,从而提高协商效率,最后给出一个实例的分析。     

AODE中的多agent协商与学习

AODE是我们研制的一个面向agent的智能系统开发环境,本文以AODE为平台研究了多agent环境下的协商与学习本文利用协商-协商过程-协商线程的概念建立了多边-多问题协商模型MMN,该协商模型支持多agent环境中的多种协商形式及agent在协商过程中的学习,系统中的学习agent采用状态概率聚类空间上的多agent强化学习算法.该算法通过使用状态聚类方法减少Q值表存储所需空间,降低了经典Q-学习算法由于使用Q值表导致的对系统计算资源的要求,且该算法仍然可以保证收敛到最优解.     

基于交错螺旋矩阵加密的自动信任协商模型

针对自动信任协商(ATN)中的敏感信息保护问题,提出了基于交错螺旋矩阵加密(ISME)的自动信任协商模型。此模型采用交错螺旋矩阵加密算法以及策略迁移法,对协商中出现的3种敏感信息进行保护。与传统的螺旋矩阵加密算法相比,交错螺旋矩阵加密算法增加了奇偶数位和三元组的概念。为了更好地应用所提模型,在该协商模型的证书中,引入了属性密钥标志位的概念,从而在二次加密时更有效地记录密钥所对应的加密敏感信息,同时列举了在协商模型中如何用加密函数对协商规则进行表示。为了提高所提模型协商成功率和效率,提出了0-1图策略校验算法。该算法利用图论中的有向图构造了6种基本命题分解规则,可以有效地确定由访问控制策略抽象而成的命题种类。之后为了证明在逻辑系统中此算法的语义概念与语法概念的等价性,进行了可靠性、完备性证明。仿真实验表明,该模型在20次协商中策略披露的平均条数比传统ATN模型少15.2条且协商成功率提高了21.7%而协商效率提高了3.6%。     

基于加速遗传模拟退火算法的协商策略研究

随着电子商务的不断发展,如何快速有效地进行自动协商是研究所面临的一个重要问题。根据Bazzar协商模型,提出了一种加速遗传模拟退火算法(AGASA),该算法将遗传算法和模拟退火算法结合,并且加入压缩搜索范围的算法加速机制,同时采用特殊的实数编码方式令算法能更加稳定地收敛。仿真试验表明,算法能快速稳定地解决协商模型所描述的协商问题。     

供应链协商调度模型与算法

研究了由一个制造商和一个分销商组成的供应链上分销商协商调度问题.此供应链中,制造商比分销商有更强的影响力,先于分销商进行调度.制造商与分销商之间不共享作业处理时间.为了改善分销商调度,建立了基于补偿的分销商协商模型,设计了保留信息私有性的协商调度策略,提出并分析了协商调度下制造商调度算法以及基于生态种群竞争的分销商协同演化调度算法.仿真实验结果表明,分销商协商调度模型与算法能够有效改善分销商调度性能,在不增加制造商调度成本的条件下,可最大程度地削减分销商调度成本超过25％.此外,提出的竞争协同演化算法能够获得比遗传算法、粒子群算法和蚁群算法更好的调度解.     

多Agent的自动协商

协商是多Agent系统实现协调、协作和冲突消解的关键环节。如何构造有效的协商模型来提高Agent的协商能力，是多Agent系统研究中待解决的问题之一。文章主要讨论了双边多项目协商问题，提出了相应的协商模型、协议和协商算法，具有一定的通用性。     

基于Nearest-Biclusters协作过滤技术的效用图结构学习算法

在多议题协商研究中,议题之间的依赖关系增加了协商Agent效用函数的复杂性,从而使得多议题协商变得更加困难.基于效用图的多议题依赖协商模型是体现议题间依赖关系的多议题协商模型.在该协商模型中,协商双方仅需要较少的协商步数就能够找到满足Pareto效率的协商结局.如何有效地学习买方Agent的效用图结构是该协商模型的关键.文中基于Nearest-Biclusters协作过滤技术的思想提出了一种新的效用图结构学习算法(NBCFL算法).该算法首先利用Nearest-Biclusters协作过滤技术发现买方偏好的局部匹配特性,提取与当前买方Agent类型相同的买方Agent所产生的协商历史记录,然后通过计算各议题间的依赖度学习买方Agent的效用图结构.实验表明在参与协商的买方Agent类型不同的条件下,NBCFL算法比IBCFL算法能更好地学习买方Agent的效用图结构.     

多Agent系统中双边多指标自动协商的ACEA算法

自动协商是多Agent系统中的一个中心议题,它是在Agent间建立一种合作合约,多数情况下这种合约包含多个协商指标,而多指标的协商比单一指标的协商要复杂得多·因此,如何快速、高效地进行Agent间的多指标自动协商是多Agent系统中必须解决的一个问题·给出了一个Agent间多指标协商的模型(MN),并在此基础上提出了双边—多指标协商的一种加速混沌进化算法(ACEA)·ACEA算法首先将混沌机制引入进化计算,然后采用压缩技术对算法进行加速,这样既克服了进化计算过早收敛到局部Nash平衡点的缺点,又解决了多指标协商繁杂的计算和引入混沌后带来的收敛速度慢的问题·理论分析和仿真实验表明,ACEA算法以概率1收敛到全局最优解·     

Efficient Management of Multi-Linked Negotiation Based on a Formalized Model

A Multi-linked negotiation problem occurs when an agent needs to negotiate with multiple other agents about different subjects (tasks, conflicts, or resource requirements), and the negotiation over one subject has influence on negotiations over other subjects. The solution of the multi-linked negotiations problem will become increasingly important for the next generation of advanced multi-agent systems. However, most current negotiation research looks only at a single negotiation and thus does not present techniques to manage and reason about multi-linked negotiations. In this paper, we first present a technique based on the use of a partial-order schedule and a measure of the schedule, called flexibility, which enables an agent to reason explicitly about the interactions among multiple negotiations. Next, we introduce a formalized model of the multi-linked negotiation problem. Based on this model, a heuristic search algorithm is developed for finding a near-optimal ordering of negotiation issues and their parameters. Using this algorithm, an agent can evaluate and compare different negotiation approaches and choose the best one. We show how an agent uses this technology to effectively manage interacting negotiation issues. Experimental work is presented which shows the efficiency of this approach.     

Classification-based self-adaptive differential evolution and its application in multi-lateral multi-issue negotiation

Multi-lateral multi-issue negotiations are the most complex realistic negotiation problems. Automated approaches have proven particularly promising for complex negotiations and previous research indicates evolutionary computation could be useful for such complex systems. To improve the efficiency of realistic multi-lateral multi-issue negotiations and avoid the requirement of complete information about negotiators, a novel negotiation model based on an improved evolutionary algorithm p-ADE is proposed. The new model includes a new multi-agent negotiation protocol and strategy which utilize p-ADE to improve the negotiation efficiency by generating more acceptable solutions with stronger suitability for all the participants. Where p-ADE is improved based on the well-known differential evolution (DE), in which a new classification-based mutation strategy DE/rand-to-best/pbest as well as a dynamic self-adaptive parameter setting strategy are proposed. Experimental results confirm the superiority of p-ADE over several state-of-the-art evolutionary optimizers. In addition, the p-ADE based multiagent negotiation model shows good performance in solving realistic multi-lateral multi-issue negotiations.     

基于博弈分析的电子商务自动协商系统

为提高基于拍卖机制的电子商务自动协商系统效率，该文以非合作博弈论为基础提出了基于博弈分析的自动协商Agent模型及基于拍卖机制和博弈分析的自动投标协商算法，实现了一个基于拍卖机制和博弈分析的电子商务自动协商原型系统，并应用在一个企业敏捷供应链管理系统中实现自动协商交易。     

Addressing stability issues in mediated complex contract negotiations for constraint-based, non-monotonic utility spaces

Negotiating contracts with multiple interdependent issues may yield non- monotonic, highly uncorrelated preference spaces for the participating agents. These scenarios are specially challenging because the complexity of the agents’ utility functions makes traditional negotiation mechanisms not applicable. There is a number of recent research lines addressing complex negotiations in uncorrelated utility spaces. However, most of them focus on overcoming the problems imposed by the complexity of the scenario, without analyzing the potential consequences of the strategic behavior of the negotiating agents in the models they propose. Analyzing the dynamics of the negotiation process when agents with different strategies interact is necessary to apply these models to real, competitive environments. Specially problematic are high price of anarchy situations, which imply that individual rationality drives the agents towards strategies which yield low individual and social welfares. In scenarios involving highly uncorrelated utility spaces, “low social welfare” usually means that the negotiations fail, and therefore high price of anarchy situations should be avoided in the negotiation mechanisms. In our previous work, we proposed an auction-based negotiation model designed for negotiations about complex contracts when highly uncorrelated, constraint-based utility spaces are involved. This paper performs a strategy analysis of this model, revealing that the approach raises stability concerns, leading to situations with a high (or even infinite) price of anarchy. In addition, a set of techniques to solve this problem are proposed, and an experimental evaluation is performed to validate the adequacy of the proposed approaches to improve the strategic stability of the negotiation process. Finally, incentive-compatibility of the model is studied.     

A novel method for coevolving PS-optimizing negotiation strategies using improved diversity controlling EDAs

In agent-mediated negotiation systems, the majority of the research focused on finding negotiation strategies for optimizing price only. However, in negotiation systems with time constraints (e.g., resource negotiations for Grid and Cloud computing), it is crucial to optimize either or both price and negotiation speed based on preferences of participants for improving efficiency and increasing utilization. To this end, this work presents the design and implementation of negotiation agents that can optimize both price and negotiation speed (for the given preference settings of these parameters) under a negotiation setting of complete information. Then, to support negotiations with incomplete information, this work deals with the problem of finding effective negotiation strategies of agents by using coevolutionary learning, which results in optimal negotiation outcomes. In the coevolutionary learning method used here, two types of estimation of distribution algorithms (EDAs) such as conventional EDAs (S-EDAs) and novel improved dynamic diversity controlling EDAs (ID²C-EDAs) were adopted for comparative studies. A series of experiments were conducted to evaluate the performance for coevolving effective negotiation strategies using the EDAs. In the experiments, each agent adopts three representative preference criteria: (1) placing more emphasis on optimizing more price, (2) placing equal emphasis on optimizing exact price and speed and (3) placing more emphasis on optimizing more speed. Experimental results demonstrate the effectiveness of the coevolutionary learning adopting ID²C-EDAs because it generally coevolved effective converged negotiation strategies (close to the optimum) while the coevolutionary learning adopting S-EDAs often failed to coevolve such strategies within a reasonable number of generations.     

Design of Roles and Protocols for Electronic Negotiations

Support for negotiations in electronic markets is one of the primary issues in today's e-commerce research. Whereas most activities are focused on automation aspects, only few efforts address the design of electronic negotiations. However, for the efficiency of electronic negotiation processes and the success of resulting settlements, it is essential to achieve an a-priori agreement among the negotiating parties about issues such as the syntax and semantics of offer specifications, the sequence of actions, possible responses, or time constraints, because these factors might influence, for instance, the fairness of the electronic negotiation.This paper demonstrates how an explicit and specific design can capture the way electronic negotiations are organised. The organisation design meta-model presented is part of SILKROAD, a design and application framework for electronic negotiations. On the basis of this framework, organisations creating an electronic market or sellers intending to offer potential buyers the option to bargain, can generate, in a flexible and efficient way, customised electronic negotiation systems supporting the roles and protocols designed. Furthermore, the consequent application of this meta-model can lead to the discovery of common negotiation patterns, eventually resulting in a reference model for electronic negotiations.     

Towards a web services and intelligent agents-based negotiation system for B2B eCommerce

With the explosive growth of the number of transactions conducted via electronic channels, there is a pressing need for the development of intelligent support tools to improve the degree and sophistication of automation for eCommerce. With reference to the BBT business model, negotiation is one of key steps for B2B eCommerce. Nevertheless, classical negotiation models are ineffective for supporting multi-agent multi-issue negotiations often encountered in eBusiness environment. The first contribution of this paper is the exploitation of Web services and intelligent agent techniques for the design and development of a distributed service discovery and negotiation system to streamline B2B eCommerce. In addition, an effective and efficient integrative negotiation mechanism is developed to conduct multi-party multi-issue negotiations for B2B eCommerce. Finally, an empirical study is conducted to evaluate our intelligent agents-based negotiation mechanism and to compare the negotiation performance of our software agents with that of their human counterparts. Our research work opens the door to the development of the next generation of intelligent system solutions to support B2B eCommerce.     

Combined Negotiations in E-Commerce: Concepts and Architecture

Combined Negotiations are a novel and general type of negotiation, in which the user is interested in many goods or services and consequently engages in many negotiations at the same time. The negotiations are independent of each other, whereas the goods or services are typically interdependent. Using currently available technology for electronic negotiations, the user conducts each negotiation separately, and has the burden of coordinating and reconciling them. The inherent complexity of combined negotiations in B2C as well as B2B e-commerce calls for software support.In our research, we aim to devise a Combined Negotiation Support System (CNSS) to help the user conduct all the negotiations at the same time. The CNSS enables the user to control and monitor the progress of the negotiations, makes sure that the specified dependencies are respected, and applies user-defined strategy rules. We have designed such a CNSS which we call CONSENSUS. The architecture of CONSENSUS relies on workflow technology, negotiating software agents, and rule engine technology. The originality of this architecture lies in the fact that the user of CONSENSUS models the combined negotiation at build time using a workflow that captures the sequencing of the individual negotiations and the dependencies between them. At runtime, software agents are assigned to individual negotiations, and they participate in the combined negotiation as actors in the workflow. The user can monitor the progress of the combined negotiation as a whole, and the progress of individual negotiations via dedicated graphical user interfaces. We rely on rule engine technology to enable the agents to use negotiation strategies.The paper introduces combined negotiations with a usage scenario. Then, combined negotiations are detailed, along with the approach taken to cope with their complexity. Afterwards, we describe the functionality a CNSS should provide, and present the architecture of CONSENSUS, together with a discussion of the underlying concepts and technologies. Furthermore, we report on our prototype implementation of CONSENSUS and illustrate it with an example. A discussion of related and future work concludes the paper.