基于认知学习的动态信任关系的度测机理研究

项目名称： 基于认知学习的动态信任关系的度测机理研究

项目编号： No.60873071

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 桂小林

作者单位： 西安交通大学

项目金额： 30万元

中文摘要： 在网格、P2P 和普适计算等各种开放式网络应用环境中，涉及数目巨大的、处在不同安全域的计算资源与软件实体，大量资源的接入，实体行为的动态变化，使得网络的各个实体之间难以建立动态的信任关系。研究和探索动态信任关系的度量与预测（简称度测）机理已成为动态信任管理技术的基础性工作，并成为急需解决的科学问题。本项目在感知网络实体行为和剖析行为间关系的基础上，研究信任关系的多维属性和心理认知特征，探索基于认知学习的动态信任关系度测机理，克服传统方法对环境特征和应用行为适应不足的问题。具体引入本体论、粗糙集、粒计算和信息熵等认知学习领域的前沿理论与方法，建立动态信任关系的形式化概念模型，研究行为感知与分析技术，提出信任度量与分配方法，改进基于机器学习的信任关系推理与决策算法，实现基于决策的信任预测与反馈。本项目的研究成果对于可信网络、可信系统和可信软件的研究具有良好的指导和借鉴作用。

中文关键词： 信任度量；上下文察觉；信任推理；认知学习；分布式系统

英文摘要： In the grid,P2P，pervasive computing and other open network environments, huge number of computing resources and software entities which are in different security domains are involved .The accession of many resources, the dynamic change of entities make various entities of network difficult to establish a dynamic trust relationship. Researching and exploring the measurement and forecast mechanism of the dynamic trust relationship has become a basis work of dynamic trust management technology ,and it has also become a scientific issue which is an urgent need to resolve. On the basis of monitoring network entities behavior and analyzing the relationship between behaviors, this project researches multi-dimensional attributes and psychological cognitive characteristics of trust relationship, explores the dynamic trust relationship of mechanism for measuring based on cognitive learning and overcomes the problem of traditional methods which is inadequate to adapt to environmental characteristic and application behavior. This project specifically introduces forefront theories and methods of cognitive learning area such as ontology, rough sets, tablets of calculation and information entropy, establishes a formal conceptual model of dynamic trust relationship, researches perception and analysis technology of behavior, proposes measurement and distribution method of trusts, improves reasoning algorithms of trust relationships based on machine learning, realizes forecasting and feedback of trust based on decision-making. The research results will have a good guidance and reference to the research on trusted network, trusted system and trusted software.

英文关键词： Trust Measurement;Context-aware; Trust Reasoning;Cognitive Learning;Distributed Systems