计算机应用 ›› 2011, Vol. 31 ›› Issue (08): 2068-2071.DOI: 10.3724/SP.J.1087.2011.02068

• 先进计算 • 上一篇    下一篇

基于社区发现的多主体信任评估

杨兴华1,2,王文杰2,王晓峰1,2,史忠植1   

  1. 1. 中国科学院计算技术研究所 智能信息处理重点实验室,北京100190
    2. 中国科学院研究生院 信息科学与工程学院,北京100049
  • 收稿日期:2011-01-24 修回日期:2011-03-25 发布日期:2011-08-01 出版日期:2011-08-01
  • 通讯作者: 杨兴华
  • 作者简介:杨兴华(1985-),男,山东泰安人,硕士研究生,主要研究方向:智能主体、可信计算;王文杰(1964-),男,陕西西安人,副教授,主要研究方向:人工智能;王晓峰(1978-),男,云南昆明人,博士研究生,主要研究方向:人工智能、数据挖掘;史忠植(1941-),男,江苏宜兴人,高级研究员,博士生导师,主要研究方向:人工智能、机器学习。
  • 基金资助:

    中国科学院研究生院院长基金资助项目(O85101JM03);国家自然科学基金资助项目(60803092);国家自然科学基金资助项目(60803092);国家自然科学基金资助项目(60803092);国家自然科学基金资助项目(60803092);国家自然科学基金资助项目(60803092);国家自然科学基金资助项目(60803092);国家973计划项目(2007CB311004);国家科技支撑项目(2006BAC08B06)

Trust evaluation based on community discovery in multi-Agent system

Xing-hua YANG1,2,Wen-jie WANG2,Xiao-feng WANG1,2,Zhong-zhi SHI1   

  1. 1. Key Laboratory of Intelligent Information Processing, Institute of Computing Technology of Chinese Academy of Sciences, Beijing 100190, China
    2. School of Information Science and Engineering, Graduate University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2011-01-24 Revised:2011-03-25 Online:2011-08-01 Published:2011-08-01
  • Contact: Xing-hua YANG

摘要: 为了解决多主体系统(MAS)的开放性、动态性和不确定性所带来的主体信任问题,提出一种基于社区发现的信任评估方法。首先使用G-N算法(GIRVAN M, NEWMAN M E J. Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the United States of America, 2002, 99(12):7821-7826)发现系统中的社区结构;然后根据推荐主体的推荐信任分别计算被评估主体的社区内部、外部声誉,进而结合直接信任形成主体的综合信任度;最后根据协作反馈实现主体信任度的动态调整。仿真实验结果表明,基于社区发现的信任评估方法能有效评估主体信任度,通过引入反馈机制能进一步提高交互成功率。

关键词: 多主体系统, 信任, 社区发现, 声誉, 反馈

Abstract: To solve the trust problem among Agents brought about by the characteristics of openness, dynamics and uncertainty of Multi-Agent System (MAS), a method for trust evaluation based on community discovery was proposed. Firstly, the G-N algorithm (GIRVAN M, NEWMAN M E J. Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the United States of America, 2002, 99(12): 7821-7826) was employed to discover the community structure in the system. Both the inner and outer community reputations of the estimated Agents were calculated respectively by use of the belief of the recommending Agents, and then the total trust value was further assessed by combining the reputations and the direct trust values. Furthermore, the dynamic adjustment of Agent's trust value was realized via cooperation feedback. Lastly, the simulation results show that the community discovery-based trust evaluation method can effectively evaluate the Agent's trust value, and further enhance the ratio of successful interactions with the introduction of the feedback mechanism.

Key words: Multi-Agent System (MAS), trust, community discovery, reputation, feedback

中图分类号: