中文摘要：

软件测试中,缺陷关联是一种比较普遍的现象。已有研究表明,充分利用缺陷之间的关联信息有助于提高软件测试效率,同时,从缺陷自身角度分析,为了综合权衡多个影响测试决策的因素（如缺陷等级、缺陷可检测率和缺陷关联关系）。提出一种基于缺陷关联的受控马尔科夫链测试模型,将测试过程转换成基于多目标权值的路径优化问题,综合权衡缺陷可检测率、关联系数和测试回扣;并运用Prim算法构造最小生成树以构造基于多目标权值的软件优化测试策略。在资源约束下,该方法利用缺陷关联引导测试决策的选取,优先检测关联紧密、可检测率大、回扣多的关联缺陷。通过仿真实验,证明了该方法的有效性。

英文摘要：

In software testing,defect correlation is a relatively common phenomenon.Studies have shown that making the best of associated information between defects is beneficial to improving software testing efficiency.Meanwhile,from the perspective of defects themselves,a controlled Markov chain model based-on the defect correlation is proposed in order to synthetically balance multiple factors influencing the decision-making（such as defect severity level,defect detecting rate and the relationships between correlated detects）.The testing is converted to route optimization problem based on multiobjective weight.Besides,defect detection rate,correlation coefficient and testing rebates are measured overall.Prim algorithm is applied to construct minimum spanning tree so as to construct software optimization testing strategy based on multi-objective weight.Under resource constraint,this method utilizes defect correlation to guide selection of testing strategy,and preferentially detect correlated defects with close correlation,large detection rate and many rebates. The simulation experiment proves effectiveness of the method.