SHF: Small: Asked and Answered: Intelligent Data Science for Software Projects

SHF：小型：询问和回答：软件项目的智能数据科学

• 批准号: 1649448
• 负责人: Jane Huang
• 金额: $ 51.59万
• 依托单位: University of Notre Dame
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-07-01 至 2020-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1649448&HistoricalAwards=false
• 关键词: SHF; Small; Asked; Answered; Intelligent

Software and systems engineering projects accrue large amounts of development data including requirements, design, code, test cases, and fault logs. When combined with the power of software analytics this data could be used to provide actionable intelligence to project stakeholders. For example, a developer might ask to view all "safety-related code which is likely to exhibit runtime faults." The proposed work will deliver a solution named Asked and Answered for Software Intensive Projects (AA) and will support a broad range of analytic queries. To foster the transition of AA to practice, the researchers will partner with industry collaborators throughout the project and develop an open-source framework facilitating the deployment of AA technology into an industrial environment. A series of natural language Query Challenges will be designed and disseminated and used to train Software Engineering students in a broad spectrum of software analytics. Delivering the AA solution requires several non-trivial challenges to be addressed. First, a natural language (NL) query interface will be developed allowing stakeholders to issue queries in their own words and from their own perspective on the project. These queries will then be transformed into a structured, executable format. Heuristics and statistical inferencing techniques will be adopted and interactive mechanisms will be designed to seek clarification from the user when the query cannot be disambiguated automatically. Software analytics will be integrated into the query mechanism so that AA can respond to a wide range of analytical questions. AA will support the dynamic composition of primitive functions into execution flows in order to service a wide range of analytical queries. Finally, AA will deliver a query engine capable of generating optimized query execution plans which take into account the nuances of the domain - namely its heterogeneous data formats, distributed tools, and the dynamic runtime requirements of analytic functions.

软件和系统工程项目积累了大量的开发数据，包括需求、设计、代码、测试用例和故障日志。当与强大的软件分析相结合时，这些数据可以用来为项目涉众提供可操作的情报。例如，开发人员可能要求查看所有“可能出现运行时错误的与安全相关的代码”。提议的工作将提供一个名为“软件密集型项目问答”（AA）的解决方案，并将支持广泛的分析查询。为了促进AA向实践的过渡，研究人员将在整个项目中与行业合作者合作，并开发一个开源框架，促进AA技术在工业环境中的部署。一系列的自然语言查询挑战将被设计和传播，并用于培训软件工程专业的学生在广泛的软件分析。交付AA解决方案需要解决几个重要的挑战。首先，将开发一个自然语言（NL）查询接口，允许涉众用他们自己的语言和从他们自己的角度对项目进行查询。然后将这些查询转换为结构化的、可执行的格式。将采用启发式和统计推理技术，并设计交互机制，以便在查询不能自动消除歧义时寻求用户的澄清。软件分析将集成到查询机制中，以便机管局可以响应广泛的分析问题。AA将支持将基本函数动态组合到执行流中，以服务于广泛的分析查询。最后，AA将提供一个能够生成优化查询执行计划的查询引擎，该查询引擎考虑到领域的细微差别——即异构数据格式、分布式工具和分析功能的动态运行时需求。

项目成果

Supporting Program Comprehension through Fast Query response in Large-Scale Systems

通过大型系统中的快速查询响应支持程序理解

• DOI: 10.1145/3387904.338926
• 发表时间: 2020
• 期刊: 2020 IEEE/ACM 28th International Conference on Program Comprehension (ICPC
• 作者: Lin, Jinfeng;Liu, Yalin;Cleland-Huang, Jane
• 通讯作者: Cleland-Huang, Jane