EAGER: Empirical Software Engineering for Computational Science

EAGER：计算科学的实证软件工程

• 批准号: 1826574
• 负责人: Timothy Menzies
• 金额: $ 12.46万
• 依托单位: North Carolina State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-05-01 至 2019-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1826574&HistoricalAwards=false
• 关键词: EAGER; Empirical; Software; Engineering; Computational

Science has become increasingly reliant on Computational Science methods implemented in software. These methods are complex, and therefore the software that implements them is prone to errors. This projects seeks to transformatively improve the state of the practice in the development of Computational Science software by applying systematic, data-driven methods (known as empirical methods) to evaluate how software is being developed and to suggest improvements. Improving the software engineering methods of Computational Science would result in higher quality software, and consequently increase our confidence in the research in scientific phenomena conducted by Computational Scientists, Much of the work in Computational Science is related to the software that implements it. In this project, the researcher will apply state of the art empirical software engineering methods to Computational Science software. Qualitative methods will be applied to conduct large scale surveys of computational science. Quantitative data mining tools (classifiers, intelligent data preprocessor, automatic hyperparameter optimizers) will be used to can learn predictive models of time series of SE data such as "Where in this system should we look for current bugs?" and "How many bugs are left on the system?". These models can be used to guide developer effort in building new code or maintaining old code.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

科学越来越依赖于在软件中实现的计算科学方法。这些方法很复杂，因此实现它们的软件容易出错。该项目旨在通过应用系统的，数据驱动的方法（称为经验方法）来评估软件的开发方式并提出改进建议，从而彻底改善计算科学软件开发的实践状态。改进计算科学的软件工程方法将产生更高质量的软件，从而增加我们对计算科学家进行科学现象研究的信心。计算科学的许多工作都与实现它的软件有关。在这个项目中，研究人员将把最先进的经验软件工程方法应用于计算科学软件。定性方法将被应用于进行计算科学的大规模调查。将使用定量数据挖掘工具（分类器、智能数据预处理器、自动超参数优化器）来学习SE数据时间序列的预测模型，例如“我们应该在该系统的何处查找当前错误？以及“系统上还有多少漏洞？“".这些模型可用于指导开发人员构建新代码或维护旧代码。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。