Deep semantic-Based Feature Envy Identification

Guo, Xueliang; Shi, Chongyang; Jiang, He

doi:10.1145/3361242.3361257

Cited by 20 publications

(11 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several Deep Learning (DL) approaches consider Feature Envy and Data Class detection problems [26] [27] [28].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Automatic detection of Feature Envy and Data Class code smells using machine learning

Škipina¹,

Slivka²,

Luburić³

et al. 2023

Preprint

View full text Add to dashboard Cite

<p> A code smell is a surface indication that usually corresponds to a deeper problem in the system. Detecting and removing code smells is crucial for sustainable software development. However, manual detection can be daunting and time-consuming. Machine learning (ML) is a promising approach towards the automation of code smell detection. The first ML-based methods were classifiers trained on feature vectors comprising software metrics extracted by off-the-shelf tools. Determining the optimal set of metrics is a complex problem that requires both ML and software engineering expertise. Recently source code embedding models emerged as a viable feature-inferring alternative. However, their potential is yet to be fully explored. To that aim, we compare state-of-the-art source code embedding models (CuBERT and CodeT5) with the models trained on metrics returned by the CK Tool and RepositoryMiner tools. We focus on detecting the Data Class and Feature Envy code smells within a large-scale, manually labeled, publicly available dataset. After extensive experiments (51 test/train splits), we found that source code embedding models have comparable performances with software metrics, a that they indeed can capture important characteristics of the source code. We discuss our findings in detail in the paper.</p>

show abstract

“…Several Deep Learning (DL) approaches consider Feature Envy and Data Class detection problems [26] [27] [28].…”

Section: Related Workmentioning

confidence: 99%

“…Guo et al [28] proposed two DL-based Feature Envy detection approaches -the Natural Language Processing-inspired models and the model that combined structural metrics and semantic (textual) features. They used the dataset created in the study by Fontana et al [16].…”

Section: Related Workmentioning

confidence: 99%

Automatic detection of Feature Envy and Data Class code smells using machine learning

Škipina¹,

Slivka²,

Luburić³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The experimental results showed that the model achieves good results on several benchmarks. Guo et al 44 proposed a deep semantic-based classifier that combines method-representation and the CNN model to detect code smells of feature envy on two feature envy data sets. They fed the metric cluster into the CNN-based model, which has three convolutional layers and does not set the pooling layer.…”

Section: Application Of Cnns In Other Fieldsmentioning

confidence: 99%

Software change‐proneness prediction based on deep learning

Zhu

Wang

2022

J Software Evolu Process

View full text Add to dashboard Cite

Software change-proneness prediction can help reduce software maintenance costs.Thus, it has drawn the attention of many researchers. In this paper, we propose a CNN (convolutional neural network)-based method for software change-proneness prediction, aiming to utilize the powerful prediction ability to make score of the performance measure higher than other baseline methods. Moreover, to alleviate the effect of the class imbalance problem, resampling methods are employed with the CNN. To validate the performance of the proposed CNN-based method, an empirical study was conducted. The experimental results show that the CNN-based method together with the resampling method performs better than the baseline methods, and the scores of performance measure of CNN with the ROS (random oversampling) method are higher than other method, especially the important performance measure.

show abstract

“…Guo et al [36] propose two DL approaches to detect Feature Envy. The first approach analyses the textual content of the method by applying DL mechanisms inspired by NLP models.…”

Section: Code Smell Detection Approachesmentioning

confidence: 99%

“…However, a significant limitation of these studies is that they do not evaluate their approaches on fully manually labeled datasets which means that the test data might contain falsely labeled instances. Furthermore, we found that only studies [36] and [39] compared their approaches to simpler (non-DL-based) alternatives.…”

Section: Code Smell Detection Approachesmentioning

confidence: 99%

Automatic detection of code smells using metrics and CodeT5 embeddings: a case study in C#

Kovačević¹,

Luburić²,

Slivka³

et al. 2022

Preprint

View full text Add to dashboard Cite

Code smells are code structures that harm the software’s quality. An obstacle to developing automatic detectors is the available datasets' limitations. Furthermore, researchers developed many solutions for Java while neglecting other programming languages. Recently, we created the code smell dataset for C# by following an annotation procedure inspired by the established annotation practices in Natural Language Processing. This paper evaluates Machine Learning (ML) code smell detection approaches on our novel dataset. We consider two feature representations to train ML models: (1) code metrics and (2) CodeT5 embeddings. This study is the first to consider the CodeT5 state-of-the-art neural source code embedding for code smell detection in C#. To prove the effectiveness of ML, we consider multiple metrics-based heuristics as alternatives. In our experiments, the best-performing approach was the ML classifier trained on code metrics (F-measure of 0.87 for Long Method and 0.91 for Large Class detection). However, the performance improvement over CodeT5 features is negligible if we consider the advantages of automatically inferring features. We showed that our model exceeds human performance and could be helpful to developers. To the best of our knowledge, this is the first study to compare the performance of automatic smell detectors against human performance.

show abstract

Deep semantic-Based Feature Envy Identification

Cited by 20 publications

References 19 publications

Automatic detection of Feature Envy and Data Class code smells using machine learning

Automatic detection of Feature Envy and Data Class code smells using machine learning

Software change‐proneness prediction based on deep learning

Automatic detection of code smells using metrics and CodeT5 embeddings: a case study in C#

Contact Info

Product

Resources

About