Coupling related code smell detection method based on deep learning

doi:10.11772/j.issn.1001-9081.2021061403

Abstract

Abstract:

Heuristic and machine learning based code smell detection methods have been proved to have limitations， and most of these methods focus on the common code smells. In order to solve these problems， a deep learning based method was proposed to detect three relatively rare code smells which are related to coupling， those are Intensive Coupling， Dispersed Coupling and Shotgun Surgery. First， the metrics of three code smells were extracted， and the obtained data were processed. Second， a deep learning model combining Convolutional Neural Network （CNN） and attention mechanism was constructed， and the introduced attention mechanism was able to assign weights to the metric features. The datasets were extracted from 21 open source projects， and the detection methods were validated in 10 open source projects and compared with CNN model. Experimental results show that the proposed model achieves the better performance with the code smell precisions of 93.61% and 99.76% for Intensive Coupling and Dispersed Coupling respectively， and the CNN model achieves the better results with the code smell precision of 98.59% for Shotgun Surgery.

Key words: code smell, coupling, deep learning, Convolutional Neural Network (CNN), attention mechanism

摘要：

基于启发式和机器学习的代码坏味检测方法已被证明具有一定的局限性，且现有的检测方法大多集中在较为常见的代码坏味上。针对这些问题，提出了一种深度学习方法来检测过紧的耦合、分散的耦合和散弹式修改这三种与耦合度相关检测较为少见的代码坏味。首先，提取三种代码坏味需要的度量并对得到的数据进行处理；之后，构建卷积神经网络（CNN）与注意力（Attention）机制相结合的深度学习模型，引入的注意力机制可以对输入的度量特征进行权重的分配。从21个开源项目中提取数据集，在10个开源项目中对检测方法进行了验证，并与CNN模型进行对比。实验结果表明：过紧的耦合和分散的耦合在所提模型中取得了更好的结果，相应代码坏味的查准率分别达到了93.61%和99.76%；而散弹式修改在CNN模型中有更好的结果，相应代码坏味查准率达到了98.59%。

关键词: 代码坏味, 耦合, 深度学习, 卷积神经网络, 注意力机制

CLC Number:

TP311

Shan SU, Yang ZHANG, Dongwen ZHANG. Coupling related code smell detection method based on deep learning[J]. Journal of Computer Applications, 2022, 42(6): 1702-1707.

苏珊, 张杨, 张冬雯. 基于深度学习的耦合度相关代码坏味检测方法[J]. 《计算机应用》唯一官方网站, 2022, 42(6): 1702-1707.

Figures/Tables 11

Fig. 1 Framework of code smell detection

Tab. 1 Metrics of code smell

度量	定义
CINT	被检测方法调用其他类中方法的数量
CDISP	判定一个方法耦合分散度的指标
CC	其他类中调用被检测方法的方法数量
CM	与被检测方法有联系的类的数量
MAXNESTING	被检测方法嵌套层次结构中最多层的层数

Fig. 2 Attention-CNN model

Fig. 3 Attention mechanism layer

Tab. 2 Projects for training set

项目名称	项目领域	NOC	NOM	LOC
argouml	UML图绘制	1 953	17 466	160 354
axion	gradle管理插件	35	313	1 096
Emmagee	性能测试工具	9	58	907
fullsync	文件同步工具	139	806	3 900
heritrix3	爬虫工具包	555	4 722	41 972
hsqldb	数据库	659	12 915	227 069
ipscan	ip端口扫描	184	933	6 584
javacc	词法分析器	180	1 487	20 861
jGroups	群组通信工具	251	1 935	13 199
jparsec	解析jQuery	237	1 274	7 387
jspwiki	Wiki系统	30	2 243	21 296
keystore	数据证书工具	272	1 593	21 215
marauroa	服务器端框架	231	1 866	19 044
picocontainer	微核心容器	1 005	6 862	48 468
quartz	分布式框架	465	4 585	41 749
QuickServer	服务器端组件	165	1 699	16 633
roller	博客服务器	549	5 040	47 848
squirrel	数据库工具	192	1 428	8 922
xalan	xslt处理器	964	10 359	188 637
xerces	xml解析器	838	10 717	142 249
you-jextractor	下载工具	76	646	2 711

Tab. 3 Training set of Intensive Coupling

方法	CINT	CDISP	MAXNESTING	INTENSIVE
exit	0.38	16	3	1
setTool	1	1	2	0
setTime	1	1	2	0
unlock	0.18	11	3	1

Tab. 4 Projects for test set

项目名称	项目领域	NOC	NOM	LOC
ArtOfIllusion	3D动画	492	6 766	103 586
FreePlane	思维导图	787	6 938	124 937
jasperreports	报表工具	2 890	23 055	202 308
Jdeodorant	代码结构分析	391	4 265	84 726
jEdit	文本编辑器	584	7 418	104 771
jfreechart	图表绘制类库	713	6 953	70 227
pmd	代码检查工具	2 194	10 105	51 296
elasticsearch	搜索服务器	3 356	31 160	198 869
netty	Java开源框架	2 458	27 183	212 343
omega	辅助翻译工具	631	4 543	39 153

Tab. 5 Results of detection of Intensive Coupling

模型	标签	查准率	查全率	$F 1$
CNN	0	100.00	99.90	99.95
CNN	1	89.44	99.73	94.30
Attention-CNN	0	100.00	99.94	99.97
Attention-CNN	1	93.61	100.00	96.70

Tab. 5 Results of detection of Intensive Coupling

模型	标签	查准率	查全率	$F 1$
CNN	0	100.00	99.90	99.95
CNN	1	89.44	99.73	94.30
Attention-CNN	0	100.00	99.94	99.97
Attention-CNN	1	93.61	100.00	96.70

Tab. 6 Results of detection of Dispersed Coupling

模型	标签	查准率	查全率	$F 1$
CNN	0	100.00	99.99	99.99
CNN	1	99.04	100.00	99.52
Attention-CNN	0	100.00	100.00	100.00
Attention-CNN	1	99.76	99.76	99.76

Tab. 6 Results of detection of Dispersed Coupling

模型	标签	查准率	查全率	$F 1$
CNN	0	100.00	99.99	99.99
CNN	1	99.04	100.00	99.52
Attention-CNN	0	100.00	100.00	100.00
Attention-CNN	1	99.76	99.76	99.76

Tab. 7 Results of detection of Shotgun Surgery

模型	标签	查准率	查全率	F₁
CNN	0	100.00	99.99	99.99
CNN	1	98.59	100.00	99.45
Attention-CNN	0	100.00	99.92	99.96
Attention-CNN	1	95.66	100.00	97.78

Tab. 8 Time consumption of three code smells

模型	过紧的耦合	分散的耦合	散弹式修改
CNN	130.49	150.85	61.21
Attention-CNN	131.68	157.60	69.58

References 23

1	MENS T， TOURWE T. A survey of software refactoring［J］. IEEE Transactions on Software Engineering， 2004， 30（2）： 126-139. 10.1109/tse.2004.1265817
2	FOWLER M， BECK K， BRANT J， et al. Refactoring： Improving the Design of Existing Code［M］. Boston： Addison-Wesley Professional， 1999： 71-76.
3	APRIL A， ABRAN A. Software Maintenance Management： Evaluation and Continuous Improvement ［M］. Hoboken： John Wiley & Sons， 2012： 1-5.
4	YOSHIDA N， KINOSHITA M， IIDA H. A cohesion metric approach to dividing source code into functional segments to improve maintainability［C］// Proceedings of the 16th European Conference on Software Maintenance and Reengineering. Piscataway： IEEE， 2012： 365-370. 10.1109/csmr.2012.45
5	PALOMBA F， BAVOTA G， DI PENTA M， et al. Mining version histories for detecting code smells［J］. IEEE Transactions on Software Engineering， 2015， 41（5）： 462-489. 10.1109/tse.2014.2372760
6	SALES V， TERRA R， MIRANDA L F， et al. Recommending move method refactorings using dependency sets［C］// Proceedings of the 20th Working Conference on Reverse Engineering. Piscataway： IEEE， 2013： 232-241. 10.1109/wcre.2013.6671298
7	MÄNTYLÄ M V， LASSENIUS C. Subjective evaluation of software evolvability using code smells： an empirical study［J］. Empirical Software Engineering， 2006， 11（3）： 395-431. 10.1007/s10664-006-9002-8
8	MAIGA A， ALI N， BHATTACHARYA N， et al. Support vector machines for anti-pattern detection［C］// Proceedings of the 27th IEEE/ACM International Conference on Automated Software Engineering. Piscataway： IEEE， 2012： 278-281. 10.1145/2351676.2351723
9	KREIMER J. Adaptive detection of design flaws［J］. Electronic Notes in Theoretical Computer Science， 2005， 141（4）： 117-136. 10.1016/j.entcs.2005.02.059
10	FONTANA F A， MÄNTYLÄ M V， ZANONI M， et al. Comparing and experimenting machine learning techniques for code smell detection［J］. Empirical Software Engineering， 2016， 21（3）： 1143-1191. 10.1007/s10664-015-9378-4
11	DI NUCCI D， PALOMBA F， TAMBURRI D A， et al. Detecting code smells using machine learning techniques： are we there yet？［C］// Proceedings of the 2018 IEEE 25th International Conference on Software Analysis， Evolution and Reengineering. Piscataway： IEEE， 2018： 612-621. 10.1109/saner.2018.8330266
12	BENGIO Y， COURVILLE A， VINCENT P. Representation learning： a review and new perspectives［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2013， 35（8）： 1798-1828. 10.1109/tpami.2013.50
13	GUO X L， SHI C Y， JIANG H. Deep semantic-based feature envy identification［C］// Proceedings of the 11th Asia-Pacific Symposium on Internetware. New York： ACM， 2019： No.19. 10.1145/3361242.3361257
14	卜依凡，刘辉，李光杰. 一种基于深度学习的上帝类检测方法［J］. 软件学报， 2019， 30（5）： 1360-1374.
	BU Y F， LIU H， LI G J. God class detection approach based on deep learning［J］. Journal of Software， 2019， 30（5）： 1360-1374.
15	KESSENTINI W， KESSENTINI M， SAHRAOUI H， et al. A cooperative parallel search-based software engineering approach for code-smells detection［J］. IEEE Transactions on Software Engineering， 2014， 40（9）： 841-861. 10.1109/tse.2014.2331057
16	FU S Z， SHEN B J. Code bad smell detection through evolutionary data mining［C］// Proceedings of the 2015 ACM/IEEE International Symposium on Empirical Software Engineering and Measurement. Piscataway： IEEE， 2015： 1-9. 10.1109/esem.2015.7321194
17	VIDAL S， VAZQUEZ H， DIAZ-PACE J A， et al. JSpIRIT： a flexible tool for the analysis of code smells［C］// Proceedings of the 34th International Conference of the Chilean Computer Science Society. Piscataway： IEEE， 2015： 1-6. 10.1109/sccc.2015.7416572
18	HADJ-KACEM M， BOUASSIDA N. A hybrid approach to detect code smells using deep learning［C］// Proceedings of the 13th International Conference on Evaluation of Novel Approaches to Software Engineering. Setúbal： SciTePress， 2018：137-146. 10.5220/0006709801370146
19	KIM D K. Finding bad code smells with neural network models［J］. International Journal of Electrical and Computer Engineering， 2017， 7（6）： 3613-3621. 10.11591/ijece.v7i6.pp3613-3621
20	DAS A K， YADAV S， DHAL S. Detecting code smells using deep learning［C］// Proceedings of the 2019 IEEE Region 10 Conference. Piscataway： IEEE， 2019： 2081-2086. 10.1109/tencon.2019.8929628
21	LANZA M， MARINESCU R. Object-Oriented Metrics in Practice： Using Software Metrics to Characterize， Evaluate， and Improve the Design of Object-Oriented Systems［M］. Berlin： Springer， 2006： 115-167.
22	CHAWLA N V， BOWYER K W， HALL L O， et al. SMOTE： synthetic minority over-sampling technique［J］. Journal of Artificial Intelligence Research， 2002， 16： 321-357. 10.1613/jair.953
23	BAHDANAU D， CHO K， BENGIO Y. Neural machine translation by jointly learning to align and translate ［EB/OL］. ［2020-12-08］. . 10.3115/v1/w14-4009

[1]	Xianfeng YANG, Jiahe ZHAO, Ziqiang LI. Text classification model combining word annotations [J]. Journal of Computer Applications, 2022, 42(5): 1317-1323.
[2]	Li’e WANG, Xiaocong LI, Hongyi LIU. News recommendation method with knowledge graph and differential privacy [J]. Journal of Computer Applications, 2022, 42(5): 1339-1346.
[3]	Hexuan HU, Huachao SUI, Qiang HU, Ye ZHANG, Zhenyun HU, Nengwu MA. Runoff forecast model based on graph attention network and dual-stage attention mechanism [J]. Journal of Computer Applications, 2022, 42(5): 1607-1615.
[4]	Yongshuai LU, Yingjie TANG, Xinran MA. Low contrast filament sizing defect detection method of non-woven fabric based on deep feature fusion [J]. Journal of Computer Applications, 2022, 42(5): 1440-1446.
[5]	Daili CHEN, Guoliang XU. Cross-domain person re-identification method based on attention mechanism with learning intra-domain variance [J]. Journal of Computer Applications, 2022, 42(5): 1391-1397.
[6]	Yi ZHUANG, Haitao ZHAO. Proposal-based aggregation network for single object tracking in 3D point cloud [J]. Journal of Computer Applications, 2022, 42(5): 1407-1416.
[7]	Zhen QU, Kunting LI, Zhixi FENG. Remote sensing image scene classification based on effective channel attention [J]. Journal of Computer Applications, 2022, 42(5): 1431-1439.
[8]	Ye ZHANG, Rong LIU, Ming LIU, Ming CHEN. Image super-resolution reconstruction network based on multi-channel attention mechanism [J]. Journal of Computer Applications, 2022, 42(5): 1563-1569.
[9]	Yongru QIU, Guangle YAO, Jie FENG, Haoyu CUI. Single image de-raining algorithm based on semi-supervised learning [J]. Journal of Computer Applications, 2022, 42(5): 1577-1582.
[10]	Wei REN, Hexiang BAI. Multi-label image classification method based on global and local label relationship [J]. Journal of Computer Applications, 2022, 42(5): 1383-1390.
[11]	Xinlin XIE, Yi XIAO, Xinying XU. Lung nodule classification algorithm based on neural network architecture search [J]. Journal of Computer Applications, 2022, 42(5): 1424-1430.
[12]	Zhihua LIU, Wenjie CHEN, Aibin CHEN. Homologous spectrogram feature fusion with self-attention mechanism for bird sound classification [J]. Journal of Computer Applications, 2022, 42(4): 1260-1268.
[13]	Junhua GU, Rui WANG, Ningning LI, Suqi ZHANG. Knowledge graph attention network fusing collaborative filtering information [J]. Journal of Computer Applications, 2022, 42(4): 1087-1092.
[14]	Yongfeng DONG, Yuehua SUN, Lichao GAO, Peng HAN, Haipeng JI. Fault diagnosis method based on improved one-dimensional convolutional and bidirectional long short-term memory neural networks [J]. Journal of Computer Applications, 2022, 42(4): 1207-1215.
[15]	Jin ZHANG, Peiqi QU, Cheng SUN, Meng LUO. Safety helmet wearing detection algorithm based on improved YOLOv5 [J]. Journal of Computer Applications, 2022, 42(4): 1292-1300.