基于数据挖掘的Web用户使用模式生成方法研究

发布时间：2018-09-05 14:51

【摘要】：测试是保证Web应用系统高质量、高性能、高可靠性的重要手段,随着Web应用系统全球范围内的广泛使用,基于Web的应用系统已经成为当前和未来软件系统开发和实施的主流。与传统的应用软件不同,它是在开放的网络中,具有广泛性、交互性和动态性等特点,但其用户行为的不可预见性及Web应用系统的复杂性,使得Web性能测试变得很困难,因此,如何构造出逼近于真实的用户使用情况是性能测试的一个主要难题。本文以Web用户使用模式生成方法和性能测试模型为主要研究内容,旨在探索如何提高性能测试的准确性。首先,对Web性能测试和Web挖掘相关的基础理论知识进行研究,在深入分析了 Web系统特征和传统性能测试过程后,发现传统的性能测试忽略了用户的真实使用情况,在测试准确性上存在不足。因此,将Web挖掘技术应用到性能测试中,提出了一种基于历史数据挖掘的用户使用模式生成方法,详细介绍了日志文件预处理,提出了频繁访问模式树生成算法,利用该算法挖掘出用户的频繁访问模式,并对用户行为特征进行分析,构造出逼近于真实使用情况的用户使用模式。其次,对传统的性能测试模型PTGM进行深入的分析后,发现该模型在测试设计阶段缺少明确的现实依据,无法准确地指导测试场景、测试用例的设计以及测试脚本的开发。因此,将用户使用模式应用到PTGM测试设计阶段,提出了一种改进的性能测试模型 WUP-PTGM(Performance Testing General Model based on Web Usage Pattren),并对该模型进行了详细阐述。最后,利用性能测试工具LoadRunner11.0对实际项目案例进行性能测试,并对性能测试结果进行评估。实验表明,本文提出的用户使用模式可以有效指导测试设计过程;利用WUP-PTGM模型指导性能测试,能够提高性能测试结果的准确性。
[Abstract]:Testing is an important means to guarantee the high quality, high performance and high reliability of Web application system. With the wide use of Web application system worldwide, the application system based on Web has become the mainstream of current and future software system development and implementation. Different from the traditional application software, it has the characteristics of extensive, interactive and dynamic in the open network, but the unpredictable user behavior and the complexity of the Web application system make it difficult to test the performance of Web. How to construct a close-to-real user usage is a major problem in performance testing. In this paper, Web user usage pattern generation method and performance test model as the main research content, to explore how to improve the accuracy of performance testing. First of all, the basic theoretical knowledge of Web performance testing and Web mining is studied. After analyzing the characteristics of Web system and the traditional performance testing process, it is found that the traditional performance testing ignores the real usage of users. There is a deficiency in the accuracy of the test. Therefore, applying Web mining technology to performance testing, a user usage pattern generation method based on historical data mining is proposed, the preprocessing of log files is introduced in detail, and a frequent access pattern tree generation algorithm is proposed. The algorithm is used to mine the frequent access patterns of users and analyze the characteristics of users' behavior to construct the user usage patterns which are close to the real usage. Secondly, after deeply analyzing the traditional performance test model (PTGM), it is found that the model lacks a clear practical basis in the test design phase, and can not accurately guide the test scenario, the design of test cases and the development of test scripts. Therefore, the user usage pattern is applied to the PTGM test design phase, and an improved performance test model (WUP-PTGM (Performance Testing General Model based on Web Usage Pattren),) is proposed and described in detail. Finally, the performance testing tool LoadRunner11.0 is used to test the performance of actual project cases, and the performance test results are evaluated. The experimental results show that the user usage mode proposed in this paper can effectively guide the test design process, and the accuracy of the performance test results can be improved by using the WUP-PTGM model to guide the performance test.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【参考文献】