基于证据理论的软件项目风险评估方法

A Software Project Risk Assessment Method Based on Evidence Theory

针对专家模糊评判语言导致风险评估结果可信性降低的问题,提出一种基于证据理论的软件项目风险评估方法。该方法首先将专家评估值转化为基本概率分配(BPA),其次使用基于相似系数优化的证据合成方法进行数据融合,并根据风险矩阵得出风险优先级。此外,相似系数可以作为复盘专家评估值相互支持情况的辅助决策依据。通过算例验证了该方法的实用性和可行性,通过与模糊理论、熵权法及TODIM方法进行对比分析,进一步验证和展现该方法在运算过程、权重分配方法以及处理冲突信息方面具备的客观性和优越性。

Aiming at the problem of reduced credibility of risk assessment results due to the fuzzy judgment language of experts,this paper proposes a software project risk assessment method based on evidence theory.Firstly,in this paper,expert assessment values are transformed into basic probability assignments(BPA),and secondly an evidence synthesis method based on similarity coefficient optimization is used for data fusion,and risk priorities based on a risk matrix are derived.In addition,the similarity coefficient can be used as an auxiliary decision basis for reviewing the mutual support of expert assessment values.The practicality and feasibility of the method are verified through the arithmetic examples,and the objectivity and superiority of the method in the arithmetic process,weight allocation method and handling of conflicting information are further verified and demonstrated through the comparative analysis with the fuzzy theory,entropy weighting method and TODIM method.