高级检索

    一种基于效用和证据理论的可信软件评估方法

    Trustworthy Software Evaluation Using Utility Based Evidence Theory

    • 摘要: 由于可信软件评估需求的动态多变以及专家主观决策的有限理性,多维多尺度可信软件评估问题是一个重要而困难的研究课题.在分析现有可信软件评估需求的基础上,提出一种基于效用和证据理论的可信软件评估方法.首先设计了一个需求驱动的可信指标树动态构造模型:开放式可信指标数据库和指标树生成算法;接着讨论分析了基于效用的可信软件定性和定量指标的信息预处理技术,并重点介绍基于分布式评估框架和Dempster合成规则的可信软件评估证据推理算法;最后通过案例证明了该方法的有效性和合理性.相信该模型的提出能对复杂环境下软件可信性评估理论的进一步研究起推动作用.

       

      Abstract: Due to the variable and dynamic characteristics in trustworthy software evaluation demands, and to the experts’ subjective decision-making with bounded rationality, to evaluate multidimensional and multi-scale trustworthy software is an important and challenging research. To solve this problem, a trustworthy software evaluation mechanism using utility based evidence theory is discussed and designed. Firstly, a dynamic construction model for trustworthy index tree is defined, including an open index node database and an index tree generation algorithm. So the model can find the correct index system which can reflect the real trustworthy evaluation demands and can be used by trustworthy evaluation agents or users. Secondly, utility theory is used to achieve quantitative and qualitative evaluation information pre-processing, namely, multi-dimension trustworthy evaluation information smooth and consistency transformation. On this basis, a trustworthy software evaluation evidential reasoning algorithm, which can analyze and solve the valuable data from the original information carried by the trustworthy evaluation experts, is given based on distributed evaluation framework and Dempster rule. Finally, the analysis and experimental results of this paper show that the validity and rationality of the proposed method is especially suitable to large scale industrial detection software. It can impel the further research of software trustworthiness evaluation theory under complicated environment.

       

    /

    返回文章
    返回