5th Chinese Search Based Software Engineering Workshop (CSBSE) 2016

第五届中国基于搜索的软件工程研讨会,黑龙江牡丹江

 

        基于搜索的软件工程(Search-Based Software Engineering, SBSE)是传统软件工程和智能计算交叉的新兴研究领域,它采用智能计算领域的现代启发式搜索优化算法解决软件工程相关问题,核心是实现智能化和自动化的软件工程相关问题求解,被2007年度IEEE国际软件工程大会(ICSE)确立为软件工程的未来发展方向之一。

        中国基于搜索的软件工程研讨会(CSBSE)旨在促进 SBSE 在国内的发展和创新。从2012年起,已经分别在北京、大连、徐州、南京成功举办了四届,第五届将于2016年6月23-24日在牡丹江师范学院召开。此次会议将聚集国内外知名专家学者,交流基于搜索的软件工程相关理论与方法的研究成果,探讨该方向的热点问题及其解决途径,本届研讨会将邀请英国伯明翰大学姚新教授等国内外知名专家作大会报告。

会议照片
报告下载

承办单位:

会议报到2016年06月23日
会议召开2016年06月24日

Programming Committee

  • Zheng Li,  Beijing University of Chemical Technology
  • Dunwei Gong,  China University of Mining and Technology
  • Changhai Nie,  Nanjing University
  • He Jiang,  Dalian University Of Technology

说明:本次会议,四位教授均在会议上做报告。

大会报告


姚新
英国伯明翰大学
个人主页

Advanced Machine Learning Approaches to Software Defect Prediction

In software defect prediction, static code attributes are extracted from previous releases of software with the log files of defects, and used to build models to predict defective modules in the next release. The prediction helps to locate parts of the software that are more likely to contain defects. This is particularly useful when the project budget is limited, or the whole software system is too large to be tested exhaustively, because a good defect predictor can guide software engineers to focus limited testing resources on defect-prone parts of the software. To achieve the best outcomes, it is essential that the predictive model is as accurate as possible. Machine learning algorithms have been used to learn such a model from historical data. This talk describes two recent efforts in applying advanced machine learning approaches to software defect prediction by exploiting the characteristics of software defect prediction problems. First, when software defect prediction is formulated as a classification problem (as many researchers do), it is a highly imbalanced problem because the number of software modules with defects is far less than that without defects. In other words, we have a lot less positive cases than the negative cases in the training data set. Such a skewed data distribution poses challenges to machine learning algorithms. We will describe a recently developed algorithm, AdaBoost.NC, which can enhance the learning performance on the minority class without sacrificing the performance on the majority class. AdaBoost.NC combines the strength of AdaBoost and negative correlation learning, two ensemble learning algorithms, in tackling the software defect prediction problem. Second, instead of formulating software defect prediction as a classification or even a regression problem, we define it as a ranking problem, because the outcome of software defect prediction is normally used to rank software modules so that testing resources could be assigned. If it is the ranking that is actually used, why not formulate software defect prediction as a learning-to-rank problem and try to learn the ranking directly? This talk will introduce the learning-to-rank approach to software defect prediction. Experimental studies using real-world software data sets have been carried out to demonstrate the strength and potential weakness of our approaches.


Xin Yao is a Chair (Professor) of Computer Science and the Director of CERCIA (Centre of Excellence for Research in Computational Intelligence and Applications) at the University of Birmingham, UK. He is an IEEE Fellow and a Distinguished Lecturer of IEEE Computational Intelligence Society (CIS). He previously served as the Editor-in-Chief (2003-08) of IEEE Transactions on Evolutionary Computation and the President (2014-15) of IEEE CIS. His main research interests include evolutionary computation, ensemble learning, and their applications, especially in software engineering. His foci on software engineering include software effort estimation, software defect prediction, software module clustering and software project scheduling. His papers won the 2001 IEEE Donald G. Fink Prize Paper Award, 2010 and 2015 IEEE Transactions on Evolutionary Computation Outstanding Paper Awards, 2010 BT Gordon Radley Award for Best Author of Innovation (Finalist), 2011 IEEE Transactions on Neural Networks Outstanding Paper Award, and many other best paper awards. He won the prestigious Royal Society Wolfson Research Merit Award in 2012 and the IEEE CIS Evolutionary Computation Pioneer Award in 2013.

赵建军
日本九州大学
个人主页

Searching Reusable Code Efficiently

Over the years of software development, a vast amount of source code has been accumulated. Many code search tools were proposed to help programmers reuse previously-written code by performing free-text queries over a large-scale codebase. Our experience shows that the accuracy of these code search tools are often unsatisfactory. One major reason is that existing tools lack of query understanding ability. In this talk, I will introduce the CodeHow, a code search technique that can recognize potential APIs a user query refers to. Having understood the potentially relevant APIs, CodeHow expands the query with the APIs and performs code retrieval by applying the Extended Boolean model, which considers the impact of both text similarity and potential APIs on code search. We evaluate CodeHow on a large-scale codebase consisting of 26K C# projects downloaded from GitHub. The experimental results shows that CodeHow outperforms conventional code search tools.


Jianjun Zhao   is currently a professor of Software Engineering at Kyushu University, Japan. Before that, he was a professor at the School of Software and the Department of Computer Science and Engineering, Shanghai Jiao Tong University, China. He also worked as a visiting scientist at the Laboratory for Computer Science, Massachusetts Institute of Technology from April 2002 to March 2003. His research interests are program analysis, code search and recommendation, and automatic programming. He has published more than 80 research papers in international journals and conferences, including ICSE, FSE, PLDI, ECOOP, ASE, ISSTA, FASE and AOSD, and has served as a PC member for international conferences such as ICSE, FSE, ECOOP, AOSD, ICSM and FASE. More information about him can be found at http://hyoka.ofc.kyushu-u.ac.jp/search/details/K006257/english.html.

刘静
西安电子科技大学
个人主页

软件模块化聚类质量评价指标

软件模块聚类问题的目的是基于软件模块间的连接关系对软件系统进行自动划分并得到模块化较好的聚类结果。随着商业和生活的需求,软件系统的功能越来越强大,同时软件系统的大小也随之上升,这使得软件模块聚类的作用尤为重要。合理的软件模块化聚类质量评价指标对于指导搜索过程寻找合适的聚类划分起着至关重要的作用。根据软件开发、维护过程中的实际需求,我们介绍了已有的评价指标、改进的策略,并提出了一种基于相似度的评价指标。该指标在解决软件系统中的全局模块和系统聚类间的单向性方面表现出良好的效果。


刘静  2004年12月在西安电子科技大学获得博士学位,之后留校任教,2009年6月破格晋升为教授。期间两次获得澳大利亚国家研究基金资助,分别在昆士兰大学和新南威尔士大学工作三年。2006年获得教育部“新世纪优秀人才支持计划”资助。2008年作为陕西省高等教育系统的唯一代表参加了全国第十届妇女代表大会。目前主要研究方向为智能计算、复杂网络与数据挖掘。已主持多项国家及省部级科研项目。2013年作为第三完成人获得国家自然科学奖二等奖,2014年获得吴文俊人工智能科学技术创新奖二等奖(个人奖)。已合作出版英文专著两部、发表国际期刊和会议论文60余篇。现为国际电气电子工程师协会(IEEE)高级会员、进化计算领域国际权威期刊《IEEE Trans. Evolutionary Computation》副编。

白晓颖
清华大学
个人主页

Large-Scale Data-Driven Testing and Search-Based Optimization

While exposing services for open access, Internet software usually needs the capability of large-scale data processing. Data-driven testing is thus necessary to detect data-sensitive defects and performance bottlenecks with a great variety of data. However, it is usually expensive to cover diversified data inputs and usage scenarios. Search-based approaches can greatly reduce the cost in searching for an optimized data set with high defect-detection potency. In this talk, I will discuss the challenges of testing at Internet scale and introduce the data-driven approach that is reinforced by search algorithms, taking the LBS (Location-Based Services) testing as an illustrative case study.


白晓颖  清华大学副教授,主要研究领域为计算机软件,长期从事软件工程的教学和科研工作。已在软件工程国内外期刊和国际会议上发表论文100余篇,作为项目负责人承担了国家自然基金、863、北京市基金、国际合作、军队预研等多项科研项目。

前期注册已截止,但欢迎相关领域的学者,研究生前来旁听特邀报告并参加讨论,欢迎各位参加 CSBSE 2016 !


请各位老师和同学在 5 月 30 日前提交注册,邀请函和回执见附件。


本次研讨会将不收取注册费。

06月23日, 星期四

世贸假日酒店一楼大厅

12:00 – 21:00报到

06月24日上午, 星期五

牡丹江师范学院国际教育学院9楼同声传译室

08:30 – 08:50开幕式、集体照
08:50 – 09:40主持:李征 教授
Advanced Machine Learning Approaches to Software Defect Prediction
姚新 教授,英国伯明翰大学
09:40 – 10:20主持:巩敦卫 教授
Searching Reusable Code Efficiently
赵建军 教授,日本九州大学
10:20 – 10:40茶歇
10:40 – 11:20主持:聂长海 教授
软件模块化聚类质量评价指标
刘静 教授,西安电子科技大学
11:20 – 12:00主持:江贺 教授
Large-Scale Data-Driven Testing and Search-Based Optimization
白晓颖 教授,清华大学
12:00 – 13:00自助午餐(牡丹江师范学院国际教育学院二楼餐厅)

06月24日下午, 星期五

牡丹江师范学院国际教育学院9楼同声传译室

13:00 – 14:15主持:刘静 教授
13:00 – 13:25
质量驱动的软件项目多目标优化调度
郑宇军 教授,浙江工业大学
13:25 – 13:50
一种融合错误检测和错误定位的测试用例集最小化技术
李征 教授,北京化工大学
13:50 – 14:15
基于测试域知识的改进遗传算法
巩敦卫 教授,中国矿业大学
14:15 – 15:30主持:姜淑娟 教授
14:15 – 14:40
覆盖表生成的演化计算方法探索
聂长海 教授,南京大学
14:40 – 15:05
面向任务的代码搜索
江贺 教授,大连理工大学
15:05 – 15:30
禁忌搜索在覆盖表生成中的应用研究
王燕 博士,南京晓庄学院
15:30 – 15:50茶歇
15:50 – 17:05主持:白晓颖 教授
15:50 – 16:15
区间系数线性双层规划问题的进化算法
李和成 教授,青海师范大学
16:15 – 16:40
一种基于遗传算法的多缺陷定位方法
樊向宇 博士,天津大学
16:40 – 17:05
基于随机理论的软件错误定位技术
王蓁蓁 教授,金陵科技学院
17:05 – 17:55主持:张岩 教授
17:05 – 17:30
GitHub开源软件的数据分析工具及应用
杨波 博士,北方工业大学
17:30 – 17:55
基于烟花爆炸优化算法的测试数据生成方法
丁蕊 博士,牡丹江师范学院
17:55 – 19:30晚餐(牡丹江师范学院国际教育学院二楼餐厅)

牡丹江师范学院国际教育学院9楼同声传译室
牡丹江,爱民区,文化街191号,牡丹江师范学院,157011


牡丹江海浪国际机场    到    会议地点

乘出租车约 17 分钟(约 9.3 公里),打车费用约 17 元

乘公共交通约 1 小时 15 分钟,票价 2 元

  • 乘坐 21路 (西二条路方向),在西二条路下车
  • 步行119米换乘 公交401路 (林业职业学院方向),在林业职业学院站下车后西步行382米

牡丹江火车站(南站)    到    会议地点

乘出租车约 17 分钟(约 6.9 公里),打车费用约 13 元

乘公共交通约 58 分钟,票价 1 元

  • 乘坐 公交12路 (师范学院方向),在师范学院站下车

牡丹江客运站    到    会议地点

乘出租车约 10 分钟(约 4.5 公里),打车费用约 9 元

乘公共交通约 41 分钟,票价 1 元

  • 乘坐 公交401路 (林业职业学院方向),在林业职业技术学院下车后西走382米

会议期间住宿统一安排在世茂假日酒店,费用需自理。

 

牡丹江世茂假日酒店

距离3.7km

牡丹江 爱民区 西地明街1号 ,近新华街。

电话: 0453-8293288

对这次会议有任何问题请联系:

张岩(Zhang yan)
Tel:0453-6511050 | 13836301123
E-mail:zhangyan@mdjnu.cn
逯强(Lu qiang)
Tel:17006699901
E-mail:mywifi@yeah.net
牡丹江师范学院(Mudanjiang Normal University, China)