动态分布环境下基于主动学习的信任模型

Trust Model Based on Active Learning in Dynamic Distributed Environment

中文关键词:  开放网络  信任管理  信任度  机器学习  交易特征

英文关键词:open network  trust management  trust value  machine learning  transaction features

中文摘要:

      随着大规模分布式系统的研究深入，动态分布环境下的信任管理成了一个突出问题。通过分析动态环境信任关系的特点，利用已有的交易模式，给出了一个分布环境的自学习信任评估模型。该模型通过对已有的交易经验特征进行学习得到各种交易模式，以此来评估新的未知主体，这解决了动态环境下新主体没有经验可寻问题。同时运用交易特征而不是具体的信任度来对主体进行信任评估，有效避免了网络中普遍存在的恶意推荐与攻击问题。

英文摘要:

      With the further research of large scale distributed systems，the trust management of dynamic distributed environment has become a prominent problem.By analyzing the characteristics of trust relationship in the dynamic environment，a self learning trust evaluation model of distributed environment with previous transactions is given.This model gains transaction patterns through learning previous transaction features.These transaction patterns are used to evaluate the new unknown users，solve the problem of new users having no experience in dynamic environment.At the same time，the use of transaction features rather than spectific trust values to assess the trustworthiness of the users，effectively avoids the prevalent malicious recommendations and attack in the network.