一种基于推荐的信誉系统初始化定值策略  

New initialization strategy of reputation system based on recommendation

作　　者：孙啸寅[1,2] 吴国新[1,2] 董永强[1,2] 叶枰[1,2] 

机构地区：[1]东南大学计算机网络和信息集成教育部重点实验室,南京210096 [2]东南大学计算机科学与工程学院,南京210096

出　　处：《东南大学学报（自然科学版）》2010年第1期41-46,共6页Journal of Southeast University：Natural Science Edition

基　　金：国家高技术研究发展计划(863计划)资助项目(2007AA01Z422)

摘　　要：为了解决在面向开放网络的信誉系统中,不合理的初始化定值策略使信誉收敛速度受阻并引发漂白攻击问题,提出一种基于推荐的信誉系统初始化定值策略.首先构建了信誉收敛形式化模型、系统动态模型和系统恶劣度模型.然后依据对乐观初始化、悲观初始化和动态初始化等初始化策略的理论分析,提出基于推荐的初始化策略,采用打分和双向匿名机制保障初始化策略的有效性和安全性.仿真结果表明,新的策略可以有效地加快信誉收敛速度和遏制漂白攻击.In the reputation system facing to the open network, unreasonable reputation initialization reduces the speed of reputation convergence and aggravates whitewashing. To solve these problems, a new initialization strategy of reputation system based on recommendation is proposed. Three mod- els including a formal model of reputation convergence, a dynamic model of system and a system model about tolerance were constructed. Following the analysis on the existent optimistic, pessimis- tic and dynamic initialization strategies, the initialization strategy of reputation system based on recommendation is proposed, which can ensure validity and secfirity by using marking and a bidirection- al anonymity communication mechanism. Simulation results show that the strategy can effectively accelerate reputation convergence and confine whitewashing.

关 键 词：信誉系统 初始化策略 推荐 漂白攻击 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]