华南理工大学学报(自然科学版) ›› 2008, Vol. 36 ›› Issue (4): 98-103.

• 计算机科学与技术 • 上一篇    下一篇

基于信任网的推荐机制

苏锦钿 郭荷清 高英   

  1. 华南理工大学 计算机科学与工程学院, 广东 广州 510640
  • 收稿日期:2007-04-23 修回日期:2007-08-07 出版日期:2008-04-25 发布日期:2008-04-25
  • 通信作者: 苏锦钿(1980-),男,博士,讲师,主要从事软件工程、Web服务组合及形式化描述、信任和声誉模型研究. E-mail:Brisk_su@hotmail.com
  • 作者简介:苏锦钿(1980-),男,博士,讲师,主要从事软件工程、Web服务组合及形式化描述、信任和声誉模型研究.
  • 基金资助:

    华南理工大学博士科研启动经费资助项目(B07-17607001Ⅱ)

Recommendation Mechanism Based on Web of Trust

Su Jin-dian  Guo He-qing  Gao Ying   

  1. School of Computer Science and Engineering, South China University of Technology, Guangzhou 510640, Guangdong, China
  • Received:2007-04-23 Revised:2007-08-07 Online:2008-04-25 Published:2008-04-25
  • Contact: 苏锦钿(1980-),男,博士,讲师,主要从事软件工程、Web服务组合及形式化描述、信任和声誉模型研究. E-mail:Brisk_su@hotmail.com
  • About author:苏锦钿(1980-),男,博士,讲师,主要从事软件工程、Web服务组合及形式化描述、信任和声誉模型研究.
  • Supported by:

    华南理工大学博士科研启动经费资助项目(B07-17607001Ⅱ)

摘要: 为深入分析信任和声誉模型中推荐机制对推荐链的依赖关系,解决推荐机制无法惩罚恶意推荐实体等问题,结合主观逻辑提出了基于信任网的推荐机制,给出了信任网的基本定义,对信任网中推荐链的依赖关系进行了形式化描述,给出了相应的解决策略,并利用信任强度解决了主观逻辑中无法对恶意推荐实体进行惩罚的问题,提高了推荐信息的精确度.模拟实验结果表明,基于信任网的推荐机制能在一定程度上减少访问到恶意实体的次数,并惩罚提供恶意推荐的实体.

关键词: 信任模型, 声誉模型, 信任网, 推荐机制, 主观逻辑, 恶意推荐

Abstract:

In order to reveal the dependence of the recommendation mechanism upon the recommendation link in trust and reputation models and to overcome the difficulty in the punishment of vicious recommendation entities, this paper proposes a recommendation mechanism based on the Web of trust by using the subjective logic, presents the basic definition of the Web of trust, gives the formal descriptions of the dependence among recommendation links,and provides the corresponding solutions. Moreover, the trust intensity is utilized to overcome the difficulty of subjective logic in the punishment of vicious recommendation entities. Thus, the precision of recommendation information is improved. Simulated results show that the recommendation mechanism based on the Web of trust can decrease the number of accessing vicious entities to some extent and effectively punish any possible vicious recommendation entities.

Key words: trust model, reputation model, Web of trust, recommendation mechanism, subjective logic, vicious recommendation