基于多agent自动协商比价系统的分析设计.pdf

摘爱摘要 Agent作为一种新型技术,其应用研究是目前人工智能发展的重要分支。协商是电子商务中的一个重要环节,协商的理论及应用研究涉及不同的学科领域。两者技术相结合称为自动协商,其应用研究对于电子商务的发展具有重要的参考意义。本文基于移动Agent技术,结合Web技术,研究构建多Agent的自动协商比价系统,用于解决电子商务中由人工协商导致的协商效率低、协商时间过长的问题。主要针对B2C模式的电子商务,解决顾客和多个商家之间就某商品而进行的一对多的多议题自动协商问题。基于对协商系统及用户需求的分析,建立了协商算法模型。作有限假设,运用提议/反提议机制进行协商,基于效用——时间制定协商策略,并引入折扣率来体现坚持——妥协策略,与效用——时间策略共同决定让步的标准。在此基础上结合电子商务实际购物模式,设计基于移动 Agent的自动协商系统框架,设计工作流程。通过原型系统——相机比价系统的开发实现,进一步验证自动协商系统在电子商务中的应用。本文提出基于移动Agent的双边多议题协商系统构建。从适应性的角度,一方面考虑建模环境更接近于现实世界,即要求协商策略的优越性;另一方面采用可移动的智能Agent,利用其在异构的软、硬件网络环境中移动的能力,及其离线计算、异步自主交互的特性,实现对同益复杂的网络环境的动态适应。工程实践中,本文采用IBM Aglet作为多Agent系统的丌发平台,利用其安全性、可扩展性等技术特点,建立了所提出的自动协商系统。采用Java语言设计并实现了具有自动协商功能的相机比价系统,实验结果对于实际应用具有指导借鉴意义。关键词:移动Agent;自动协商;Aglet 北￡iT、lkJj’≯TfI;t埘!I’。≯f一论殳 Abstract As a new technology,agent has been the significant branch of Artificial Intelligence in recent is akey part merce,theory and application about it are involved indifferent science bination ofthem is named automated application has great reference value in the development merce. With constructing automated parative—price system based on mobile bined with webtechnology,problems oflowefficiency and long timecaused by manual negotiation merce could B2C pattern merce,automated negotiation ofone—to—many muti-item about some merchandise between one customer and some merchants was concentrated on. According to analysis on negotiation system and consumer demand,negotiation algorithm model was limited assumption,offer/counter offer Was used was decided based on effectiveness- rate was introduced toembody strategy of promise,bined strategy ofeffectiveness—time todetermine standard tothe factshopping pattern ofauto-negotiation system based on mobile agent was designed,and SO prototype system,web site ofcamera parision Wasdesigned andrealized,application ofautomated negotiationsystem merce was proof-te