一种基于推荐偏差的多主体系统信任推荐模型  

Deviation Based Trust Recommendation Model for Multi-agent System

作　　者：张大鹏[1,2] 左孟飞 王新生[1] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004 [2]桂林电子科技大学广西可信软件重点实验室,广西桂林541004

出　　处：《小型微型计算机系统》2014年第8期1852-1855,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61303129)资助;广西可信软件重点实验室开放基金项目(KX201212)资助

摘　　要：已有的信任推荐模型认为推荐信任会沿传递路径衰减,这并不完全符合人类的信任习惯.针对这种情况,提出一种基于推荐偏差的多主体系统信任推荐模型:(1)对推荐信任沿传递路径衰减的不足之处进行了阐释并给出了推荐偏差的概念;(2)给出了利用推荐偏差计算推荐信任度的方法,该方法引入了信心指数来衡量被推荐者对推荐者以固定的推荐偏差向其进行推荐的信心;(3)给出了信任信息的传递方法;(4)使用了4组数据进行了场景分析,分析结果表明:当信任网络中存在信心指数较高的推荐者时,使用本方法得到的结果准确度较高.Current trust recommendation models believe that recommended trust would decrease along its path, while it doesn＇t com- pletely accord with the trust habit of human. To solve it, a deviation based trust recommendation model for multi-agent system is pro- posed. Firstly, the defect of decrement of recommended trust is discussed and the concept of recommendation deviation is proposed. Secondly, recommended trust computation method is proposed, which uses confidence factor to show the trustee＇s confidence about that the trustor will recommend with fixed deviation. Thirdly, the transferring method of recommended trust is proposed. Finally, scene analysis with four groups of data shows that the accuracy of our method is good when there are trustors with high confidence factor in the system.

关 键 词：信任推荐 多主体系统 推荐偏差 信心指数 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]