【作者】 高亚宁；

【导师】 孙文辉；

【作者基本信息】 北京交通大学 ， 计算机技术领域， 2011， 硕士

【摘要】 随着信息技术的飞速发展,软件系统在经济、金融、医疗、通信、交通、航天、航空、工业控制等领域得到了广泛的应用。因此软件的可靠性也越来越受到人们的重视。为了提高软件的可靠性,人们采用软件工程学方法来指导整个软件研发流程。在软件开发的整个过程中,在软件的编码过程中往往存在着大量的克隆代码,它们差别细微,分散在软件的不同部分,手工方法难以进行统一维护,在测试阶段中的代码审查阶段不容易对它们进行检测,这是软件测试的一个难点。在使用测试用例对程序进行测试的过程中用手工方式生成测试数据的方法工作量大、效率低、容易出错,而且无法保证测试的充分性。因此,如何生成测试数据也成为软件测试的难点之一。数据挖掘是涉及数据库、人工智能、数理统计、机器学习等的交叉学科。它可以发现大量数据间的隐含的关系。在本论文中作者研究了数据挖掘的技术与软件测试的特点,将两者结合起来来解决上述中软件测试中遇到的难点。针对代码克隆的现象,本文提出一种新的解决办法,一种数量化的方法来进行代码克隆的查找；针对手工编制结构测试数据极为困难且效率低下这一问题,根据其需要大量测试用例的特点,本文中使用遗传算法的方法来完测试用例的生成并自动完成其测试过程。更多还原

【Abstract】 With the rapid development of the information technology, software systems in the economic, financial, medical, communications, transportation, aerospace, aviation, industrial control and other fields has been widely used. Therefore, the reliability of the software more and more attention has been paid.In order to improve software reliability, software engineering methodology has been applied to guide software development procedure. In the whole process of software development, software coding process in the presence of large numbers of clones are often the code, they differ slightly, scattered in different parts of the software, manual methods are difficult to maintain a unified, in the testing phase of the code review stage is not easy detect them, this is a difficult software testing. In the use of test cases to test the program during the test data generated by hand, heavy workload, low efficiency, error-prone, and can not guarantee the adequacy of the test. Therefore, how to generate test data has also become one of the difficulties of software testing.Data mining is related to databases, artificial intelligence, statistics, machine learning and other cross-disciplinary. It can be found hidden among large amounts of data relationships. In this paper, In this paper, investigated the effects of data mining technology and software testing features, and combining the two in software testing to solve the difficulties encountered.For the phenomenon of cloning for the code, this paper presents a new solution, a quantitative approach to the search for code clones; structure for the preparation of test data by hand is extremely difficult and inefficient this problem, a large number of test cases according to their needs characteristics of genetic algorithms used in this way to complete the generation of test cases and automate the testing process.更多还原