基于改进神经网络的静态软件缺陷自动分配方法  

Static Software Defect Automatic Allocation Method Based on Improved Neural Network

作　　者：马辉[1] MA Hui(Henan Vocational College of Quality Engineering,Pingdingshan Henan 467000,China)

机构地区：[1]河南质量工程职业学院,河南平顶山467000

出　　处：《信息与电脑》2021年第4期68-70,共3页Information & Computer

摘　　要：针对软件缺陷分配存在的问题,笔者设计了基于改进神经网络的静态软件缺陷自动分配方法,首先利用隐含语义分析方法对缺陷数据进行分解,然后利用初始标签对改进神经网络进行训练,最后采用训练后的神经网络对数据进行标注,将标注后的缺陷数据利用one hot向量分配给对应的修复人员,完成自动分配。实验结果证明,设计方法的分配准确率优于传统方法,证明设计方法具有可行性。Aiming at the problems of software defect allocation,the author designed a static software defect automatic allocation method based on improved neural network.First,the defect data is decomposed by the implicit semantic analysis method,then the improved neural network is trained by the initial label,and finally the training is used.The latter neural network annotates the data,and uses the one hot vector to assign the annotated defect data to the corresponding repair personnel to complete the automatic assignment.The experimental results prove that the allocation accuracy of the design method is better than that of the traditional method,which proves the feasibility of the design method.

关 键 词：分配准确率 自动分配 隐含语义分析 神经网络 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]