基于语义相似度和进化优化的软件变异测试理论与方法

Mutation testing is a fault-based software testing technique that receives growing interest. Yet, it is still perceived as being costly and impractical. This remains a barrier to wider uptake within actual software testing. Considering the existing problems of mutation testing, this project mainly researches the theories and methods of mutation testing based on semantic similarity and evolutionary optimization. By research, we plan to give the measurement of semantic similarity between the mutated statement and the original one, laying a foundation for mutant operator reduction: propose the theory and evolutionary solution of mutant operator reduction based on semantic Mutation testing is a fault-based software testing technique that receives growing interest. Yet, it is still perceived as being costly and impractical. This remains a barrier to wider application within actual software testing. Considering the existing problems of mutation testing, this project mainly researches the theories and methods of mutation testing based on semantic similarity and evolutionary optimization. By research, we plan to give the measurement of semantic similarity between the mutated statement and the original one, laying a foundation for mutant operator reduction: propose the theory and evolutionary solution of mutant operator reduction based on semantic similarity, greatly reducing the number of mutant operators; establish the usage model of generating test data for statistical mutation testing, and propose corresponding evolutionary optimization solutions, further improving the efficiency of test data generation; finally, apply the proposed theories and methods to actual software testing, effectively improve the quality of software. This project is an innovative and challenging direction by integrating the knowledge of automation, computer, and applied mathematics, which has not only definite application background and industry demand, also has stark novelty and challenge. The expected achievements can further enrich the theory of mutation testing, improve the efficiency of mutation testing, and expand the application of evolutionary optimization. Therefore, this project has important theoretical significance and practical value.

变异测试是一种面向缺陷检测的软件测试技术，近年来引起人们越来越高的兴趣。但是，传统变异测试方法需要消耗大量计算资源，很难在实际测试中得以应用。本项目针对变异测试存在的困难问题，研究基于语义相似度和进化优化的变异测试理论与方法。通过研究，期望给出语义相似度的度量方法，为约简变异算子奠定基础；提出基于语义相似度的变异算子约简理论和进化求解方法，大幅度减少变异算子的数量；建立统计变异测试数据生成问题的使用模型，设计基于进化优化的统计变异测试数据自动生成方法，进一步提高测试数据生成的效率；最后，将上述理论与方法应用于实际的软件变异测试,有效提高软件的质量。本项目是自动化、计算机与应用数学等多个学科有机交叉的研究方向，不仅有明确的应用背景和产业需求，还具有鲜明的新颖性与挑战性。产生的研究成果能够进一步丰富变异测试理论，提高变异测试效率，拓展进化优化的应用范围，因此，具有重要的理论意义和实用价值。

变异测试是一种面向缺陷检测的软件测试技术，近年来引起人们越来越高的兴趣。但是，传统变异测试方法需要消耗大量计算资源，很难在实际测试中得以应用。项目组成员针对变异测试存在的困难问题，进行了为期4年的深入研究，给出了变异体语义相似度的度量方法，提出了基于语义相似度的变异算子约简理论和进化求解方法，建立了统计变异测试数据生成问题的使用模型，设计了基于进化优化的统计变异测试数据自动生成方法，最后，将上述理论与方法应用于实际的软件变异测试。.基于上述成果，获江苏省科学技术奖自然科学二等奖1项；申请发明专利13项，其中，已授权8项；授权登记计算机软件著作权2件；在被SCI或EI等检索的学术期刊或会议上发表论文14篇，其中，SCI源刊论文6篇，中科院核心期刊5篇；培养博士研究生2名、硕士研究生4名，圆满完成了项目的预期目标。.本项目是自动化、计算机与应用数学等多个学科有机交叉的研究方向，不仅有明确的应用背景和产业需求，还具有鲜明的新颖性与挑战性。产生的研究成果进一步丰富了变异测试理论，提高了变异测试效率，拓展了进化优化的应用范围，因此，具有重要的理论意义和实用价值。