CoRT: Transformer-based code representations with self-supervision by predicting reserved words for code smell detection

Empirical Software Engineering, Journal Year: 2024, Volume and Issue: 29(3)

Published: April 8, 2024

Language: Английский

---

Machine Learning-Based Methods for Code Smell Detection: A Survey

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(14), P. 6149 - 6149

Published: July 15, 2024

Code smells are early warning signs of potential issues in software quality. Various techniques used code smell detection, including the Bayesian approach, rule-based automatic antipattern identification utilizing B-splines, Support Vector Machine direct, SMURF (Support Machines for design detection using relevant feedback), and immune-based strategy. learning (ML) has taken a great stride this area. This study includes studies applying ML algorithms from 2005 to 2024 comprehensive manner survey provide insight regarding smell, frequently applied, metrics. Forty-two pertinent allow us assess efficacy on selected datasets. After evaluating various based open-source project datasets, evaluated additional threats obstacles such as lack standardized definitions, difficulty feature selection, challenges handling large-scale The current only considered few factors identifying smells, while study, several contributing included. Several examined, approaches, dataset languages, metrics presented. provides produce better results fills gap body knowledge by providing class-wise distributions algorithms. Machine, J48, Naive Bayes, Random Forest models most common detecting smells. Researchers can find helpful anticipating taking care development implementation issues. findings which highlight practical implications quality improvement, will help engineers fix problems during ensure

Language: Английский

---

Ensemble methods with feature selection and data balancing for improved code smells classification performance

Engineering Applications of Artificial Intelligence, Journal Year: 2024, Volume and Issue: 139, P. 109527 - 109527

Published: Oct. 28, 2024

Language: Английский

---

MARS: Detecting brain class/method code smell based on metric–attention mechanism and residual network

Journal of Software Evolution and Process, Journal Year: 2021, Volume and Issue: 36(1)

Published: Nov. 4, 2021

Abstract Code smell is the structural design defect that makes programs difficult to understand, maintain, and evolve. Existing works of code detection mainly focus on prevalent smells, such as feature envy, god class, long method. Few have been done detecting brain class/method. Furthermore, existing deep‐learning‐based approaches leverage CNN model improve accuracy by barely increasing number layers, which may cause a problem gradient degradation. To this end, paper proposes novel approach called MARS detect improves degradation employing an improved residual network. It increases weight value those important metrics label smelly samples introducing metric–attention mechanism. support training MARS, dataset BrainCode generated extracting more than 270,000 from 20 real‐world applications. evaluated compared other machine‐learning‐based approaches. The experimental results demonstrate average 2.01 % higher approaches, state‐of‐the‐art.

Language: Английский

---

Python code smells detection using conventional machine learning models

PeerJ Computer Science, Journal Year: 2023, Volume and Issue: 9, P. e1370 - e1370

Published: May 29, 2023

Code smells are poor code design or implementation that affect the maintenance process and reduce software quality. Therefore, smell detection is important in building. Recent studies utilized machine learning algorithms for detection. However, most of these focused on using Java programming language datasets. This article proposes a Python dataset Large Class Long Method smells. The built contains 1,000 samples each smell, with 18 features extracted from source code. Furthermore, we investigated performance six models as baselines were evaluated based Accuracy Matthews correlation coefficient (MCC) measures. Results indicate superiority Random Forest ensemble by achieving highest 0.77 MCC rate, while decision tree was best performing model Rate 0.89.

Language: Английский

---

Code Smell Detection Using Deep Learning Models to Enhance the Software Quality

Communications in computer and information science, Journal Year: 2025, Volume and Issue: unknown, P. 31 - 43

Published: Jan. 1, 2025

Language: Английский

---

A Semisupervised Learning Approach for Code Smell Detection

SN Computer Science, Journal Year: 2025, Volume and Issue: 6(2)

Published: Feb. 6, 2025

Language: Английский

---

Graph Neural Network-based Long Method and Blob Code Smell Detection

Science of Computer Programming, Journal Year: 2025, Volume and Issue: unknown, P. 103284 - 103284

Published: Feb. 1, 2025

Language: Английский

---

Bmco-o: a smart code smell detection method based on co-occurrences

Automated Software Engineering, Journal Year: 2025, Volume and Issue: 32(1)

Published: Feb. 21, 2025

Language: Английский

---

An Empirical Study on Feature Extraction Using Ensemble Deep Learning for Code Smell Detection

Smart innovation, systems and technologies, Journal Year: 2025, Volume and Issue: unknown, P. 573 - 585

Published: Jan. 1, 2025

Language: Английский

---