Intelligent Code Quality Management for Software Evolution

软件应用在我们的日常生活中无处不在。我们依靠他们来支付账单、购物和在线视频流媒体。它们的质量是至关重要的，不能因其不断增长的用户基础和编程复杂性而受到影响。为了确保软件应用程序的质量，提供了软件分析来自动分析收集的软件工程数据(例如，提交日志)并构建预测模型，以帮助开发人员提前识别缺陷并提高软件质量。然而，由于部署了许多大规模和复杂的应用程序，现有的软件分析方法具有以下限制：(1)缺乏行业适应性，因为以前的研究不能很好地扩展以处理不断增加的软件操作数据(例如，崩溃报告)；以及(2)很少考虑最终用户体验到的软件的感知质量，并平等对待所有缺陷。在拟议的研究中，我们的目标是在以下目标中解决上述限制。O1)为处理不断增长的代码规模提供智能支持，以交付高质量的软件。我们将开发前沿的方法和工具，以实现以下目标：O1.1)提供智能克隆管理基础设施，该基础设施可以跟踪克隆进化，并基于需要更多关注的各种标准(例如，故障倾向性)对克隆进行排名；O1.2)提取可重用组件作为应用程序编程接口(API)，以重用通常嵌入在克隆代码中的功能；以及O1.3)识别并推荐API使用模式，以演示如何实现特定功能。O2)通过识别影响最大的缺陷来提高软件的感知质量。鉴于软件应用程序的庞大和广泛的用户基础，需要新的质量改进过程和方法，以确保所开发的软件具有高质量。我们将：O2.1)提出感知质量(即用户体验和满意度)的概念，以确定要首先修复的高影响缺陷；以及O2.2)优化发布前工作的时间安排，以确保软件及时发布，但高影响缺陷在测试期间被捕获。建议的研究重点是开发解决方案，并在工业环境中通过实证研究进行测试。由于我的研究的实际影响，私营部门公司(例如IBM)和开放源码社区(例如Mozilla基金会)对采用我们的研究成果表现出了强烈的兴趣。拟议研究的结果将帮助实践者以高效的方式交付可靠和理想的软件。将感知质量整合到演进过程中，帮助从业者专注于可以快速提高感知质量的关键问题，从而提高客户满意度、品牌声誉，并最终增加公司收入。拟议的研究将培养3名博士生和2名硕士研究生。

Software applications are omnipresent in our daily lives. We rely on them to pay our bills, to shop, and to stream videos online. Their quality is critical and cannot be compromised by their ever-increasing user base and programming complexity. To ensure the quality of software applications, software analytics are provided to automatically analyze the collected software engineering data (e.g., commit logs) and build predictive models to help developers identify defects in advance and improve software quality. However, as many large-scale and complex applications are deployed, existing software analytics approaches have the following limitations: (1) lack of industrial adaption as prior research cannot scale well to handle the increasing volume of software operational data (e.g., crash reports); and (2) rare considerations of the perceived quality of software experienced by end-users and treat all defects equal. In the proposed research, we aim to address the above limitations in the following objectives. O1) Providing intelligent support for handling the increasing scale of the code to deliver high quality software. We will develop leading-edge methods and tools to O1.1) provide smart clone management infrastructure that can track clone evolution and rank the clones based on various criteria (e.g., the fault-proneness) that require more attention; O1.2) extract reusable components as application programming interfaces (APIs) to reuse the functionality that are often embedded in the cloned code; and O1.3) identify and recommend API usage patterns to demonstrate how to implement a certain functionality. O2) Improving perceived quality of software by identifying the most impactful defects. Given the large and wide user base of software applications, new quality improvement processes and methodologies are needed ensuring that the developed software is of high quality. We will: O2.1) propose the notion of perceived quality (i.e., user experience and satisfaction) to identify high impact defects to fix first; and O2.2) optimize timing of pre-release efforts to ensure that software is released in a timely fashion, but the high impact defects are captured during the testing period. The proposed research focus is the development of solutions that are tested through empirical studies in an industrial setting. Due to the practical impact of my research, private sector companies (e.g., IBM) and open-source communities (e.g., the Mozilla Foundation) have shown strong interest in adopting our research results. The outcome of the proposed research will help practitioners deliver reliable and desirable software in an efficient manner. Integrating perceived quality into the evolution process help practitioners focus on the critical issues that can quickly improve perceived quality, thereby increasing customer satisfaction, brand reputation and ultimately company revenues. The proposed research will train 3 PhD and 2 MSc students.