Finding Bad Code Smells with Neural Network Models

Kim, Dong Kwan

doi:10.11591/ijece.v7i6.pp3613-3621

Cited by 17 publications

(15 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are currently several literature reviews and surveys in the field of code smell detection and refactoring [10]. We have found in the literature that there are many studies that have provided different approaches and strategies for detecting bad code smells in modern software systems [5], [11]- [13]. Kim [5] proposed a system based on a neural network model for detecting bad cods smells and clarifies the relevance between code smells and object-oriented metrics.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Deep convolutional neural network model for bad code smells detection based on oversampling method

Khleel

Nehéz²

2022

IJEECS

View full text Add to dashboard Cite

Code <span>smells refers to any symptoms or anomalies in the source code that shows violation of design principles or implementation. Early detection of bad code smells improves software quality. Nowadays several artificial neural network (ANN) models have been used for different topics in software engineering: software defect prediction, software vulnerability detection, and code clone detection. It is not necessary to know the source of the data when using ANN models but require large training sets. Data imbalance is the main challenge of artificial intelligence techniques in detecting the code smells. To overcome these challenges, the objective of this study is to presents deep convolutional neural network (D-CNN) model with synthetic minority over-sampling technique (SMOTE) to detect bad code smells based on a set of Java projects. We considered four code-smell datasets which are God class, data class, feature envy and long method and the results were compared based on different performance measures. Experimental results show that the proposed model with oversampling techniques can provide better performance for code smells detection and prediction results can be further improved when the model is trained with more datasets. Moreover, more epochs and hidden layers help increase the accuracy of the model.</span>

show abstract

Section: Related Workmentioning

confidence: 99%

“…Several static analysis tools and code restructuring methods have been developed to discover and solve source code problems, and these tools and methods provide various ways of analyzing source codes [4]. Previous studies have classified code smells into three main categories: application, class , and method level smells [5].…”

Section: Introductionmentioning

confidence: 99%

Deep convolutional neural network model for bad code smells detection based on oversampling method

Khleel

Nehéz²

2022

IJEECS

View full text Add to dashboard Cite

show abstract

“…Cohesion Previous studies 47,48,50,54,55,57,59,65,69,70,72,73,87,89,90,[97][98][99]101,102,104,105,107,[110][111][112]115,116,120,121,124,125,128,132,164,165,179 Size Previous studies 45,[48][49][50][54][55][56][57]59,65,69,70,87,[98][99]…”

Section: Quality Attributes Referencesmentioning

confidence: 99%

Software smell detection techniques: A systematic literature review

AbuHassan

Alshayeb

Ghouti

2020

J Software Evolu Process

View full text Add to dashboard Cite

Software smells indicate design or code issues that might degrade the evolution and maintenance of software systems. Detecting and identifying these issues are challenging tasks. This paper explores, identifies, and analyzes the existing software smell detection techniques at design and code levels. We carried out a systematic literature review (SLR) to identify and collect 145 primary studies related to smell detection in software design and code. Based on these studies, we address several questions related to the analysis of the existing smell detection techniques in terms of abstraction level (design or code), targeted smells, used metrics, implementation, and validation. Our analysis identified several detection techniques categories. We observed that 57% of the studies did not use any performance measures, 41% of them omitted details on the targeted programing language, and the detection techniques were not validated in 14% of these studies. With respect to the abstraction level, only 18% of the studies addressed bad smell detection at the design level. This low coverage urges for more focus on bad smell detection at the design level to handle them at early stages. Finally, our SLR brings to the attention of the research community several opportunities for future research.

show abstract

“…On the other hand, software data mining studies which are based on sofware code and code change artifacts are common [9]. For instance, code smells in the source code have been investigated using Neural Network Models in a recent study [11]. We observe that using CFP data and data mining for Educational purposes is even more rare in the literature.…”

Section: Introductionmentioning

confidence: 97%

Using Data Mining to Identify COSMIC Function Point Measurement Competence

Bağrıyanık

Karahoca

2018

IJECE

View full text Add to dashboard Cite

Cosmic Function Point (CFP) measurement errors leads budget, schedule and quality problems in software projects. Therefore, it’s important to identify and plan requirements engineers’ CFP training need quickly and correctly. The purpose of this paper is to identify software requirements engineers’ COSMIC Function Point measurement competence development need by using machine learning algorithms and requirements artifacts created by engineers. Used artifacts have been provided by a large service and technology company ecosystem in Telco. First, feature set has been extracted from the requirements model at hand. To do the data preparation for educational data mining, requirements and COSMIC Function Point (CFP) audit documents have been converted into CFP data set based on the designed feature set. This data set has been used to train and test the machine learning models by designing two different experiment settings to reach statistically significant results. Ten different machine learning algorithms have been used. Finally, algorithm performances have been compared with a baseline and each other to find the best performing models on this data set. In conclusion, REPTree, OneR, and Support Vector Machines (SVM) with Sequential Minimal Optimization (SMO) algorithms achieved top performance in forecasting requirements engineers’ CFP training need.

show abstract

Finding Bad Code Smells with Neural Network Models

Cited by 17 publications

References 12 publications

Deep convolutional neural network model for bad code smells detection based on oversampling method

Deep convolutional neural network model for bad code smells detection based on oversampling method

Software smell detection techniques: A systematic literature review

Using Data Mining to Identify COSMIC Function Point Measurement Competence

Contact Info

Product

Resources

About