• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2013, Vol. 35 ›› Issue (9): 117-121.

• 论文 • 上一篇    下一篇

基于多Agent学习机制的服务组合

赵海燕1,2,曹健3,徐文博3   

  1. (1.上海理工大学光电信息与计算机工程学院,上海 200093;2.上海现代光学系统重点实验室,上海 200093;
    3.上海交通大学计算机科学与技术系,上海 200030)
  • 收稿日期:2013-04-30 修回日期:2013-07-25 出版日期:2013-09-25 发布日期:2013-09-25
  • 基金资助:

    国家自然科学基金资助项目(61073021,61272438,60970012);上海市科委项目(12511502704,11511500102,10DZ1200200);上海交通大学医工交叉项目(YG2011MS38);上海市教委科研创新项目(13ZZ112)

Service composition based on multiagent learning mechanism  

ZHAO Haiyan1,2,CAO Jian3,XU Wenbo3   

  1. (1.School of OpticalElectrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093;
    2.Shanghai’s Key Laboratory of Modern Optical System,Shanghai 200093;
    3.Department of Computer Science and Technology,Shanghai Jiao Tong University,Shanghai 200030,China)
  • Received:2013-04-30 Revised:2013-07-25 Online:2013-09-25 Published:2013-09-25

摘要:

服务Agent能够利用本身的功能以及来自外部的多个软件服务,向外界提供更为高层的服务。为了向用户提供综合的功能,服务Agent之间必须通过相互协同实现服务组合。提出了一种依靠多Agent学习机制进行服务组合的方法,其独特性在于通过学习,这些Agent能够对类似的输入迅速得到优化解。对基于多Agent协同进行服务组合的框架进行了介绍,在此基础上详细说明了如何利用多Agent学习机制进行服务组合,并对相关的实验进行了介绍。

关键词: 服务组合, 服务Agent, 多Agent学习

Abstract:

Service agent can make use of its own functions and external software services to provide the outside with highlevel services. In order to offer more synthetic functions, service agents should cooperate with each other. A service composition approach based on multiagent learning mechanism is proposed. Its unique feature is that, through learning, agents can improve the speed of solving similar problems after several runs of learning. Firstly, the framework for service composition based on service agents is introduced. Secondly, the multiagent learning mechanism is explained in detail. Finally, the experiment and its results are also given.

Key words: service composition;service agent;multiagent learning