基于大数据的开源项目 缺陷报告智能预检技术  

Big-data based intelligent bug triage techniques for open-source projects

作　　者：席圣渠 徐锋[1,2] 陈鑫 李宣东[1,2] XI Shengqu;XU Feng;CHEN Xin;LI Xuandong(Department of Computer Science,Nanjing University,Nanjing 210023,China;State Key Laboratory for Novel Software Technology at Nanjing University,Nanjing 210023,China)

机构地区：[1]南京大学计算机科学与技术系,江苏南京210023 [2]计算机软件新技术国家重点实验室(南京大学),江苏南京210023

出　　处：《大数据》2021年第1期48-63,共16页Big Data Research

基　　金：国家重点研发计划基金资助项目(No.2016YFB1000800);国家自然科学基金资助项目(No.61672274,No.61702252);江苏省协同创新中心项目。

摘　　要：缺陷报告预检目标在于确定优先级和修复措施,是保障软件可信的关键环节。然而,在日益普及的开源项目中,由于缺陷数量众多、缺乏组织管理等特性,人工预检难以及时完成,迫切需要基于大数据的自动化、智能化预检技术。结合工业界、学术界对缺陷报告预检的认知,提出了一种智能化缺陷报告预检技术框架,全面系统地归纳了缺陷报告预检中存在的3个关键任务:缺陷优先级分类、缺陷分派、缺陷再分派,并结合开源项目的特点提出了相关技术。实验结果初步验证了上述技术的合理性和有效性。Bug triage aims to determine the priority and repair measures and is critical in ensuring software trustability.However,in the increasingly popular open-source projects,due to a large number of defects and lack of organization and management,it is challenging to triage all the bug reports by hand on time,making big-data based,automated and intelligent bug triage urgent.An intelligent bug triage technical framework based on industry and academia’s cognition was proposed,and three key tasks:bug priority classification,bug assignment,and bug reassignment,were identified comprehensively and systematically.Related technologies for the characteristics of open-source projects were proposed.The preliminary experiment results show the reasonableness and effectiveness of the above techniques.

关 键 词：缺陷报告预检 缺陷优先级 缺陷分派 缺陷再分派 

分 类 号：G434[文化科学—教育技术学]