Software Defect Prediction Based on Effective Fusion of Multiple Features

In the field of software engineering, the goal of software defect prediction is to accelerate the detection of potential defective modules in products, thereby enhancing the efficiency of allocating testing resources effectively. Traditional software defect prediction models typically rely on a sing...

Full description

Saved in:
Bibliographic Details
Main Authors: Chaozheng Zhang, Junhua Wu
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10549525/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850271845453922304
author Chaozheng Zhang
Junhua Wu
author_facet Chaozheng Zhang
Junhua Wu
author_sort Chaozheng Zhang
collection DOAJ
description In the field of software engineering, the goal of software defect prediction is to accelerate the detection of potential defective modules in products, thereby enhancing the efficiency of allocating testing resources effectively. Traditional software defect prediction models typically rely on a single feature or perform simple aggregation of multiple features during training. However, the former overlooks the potential contributions of other features in defect prediction, while the latter fails to fully exploit the value of different features in the model. Therefore, this paper proposes a software defect prediction model based on the effective fusion of multiple features, named DP-SSCT (Semantics, Source Code, and Tradition). The model integrates traditional features, source code features, and semantic features. To determine the importance of each feature in the model, a gate fusion mechanism and attention mechanism are introduced to capture significant information in the fused features. The proposed model is evaluated on nine open-source Java project samples from the PROMISE dataset. Experimental results demonstrate that the model outperforms existing prediction models, exhibiting superior performance and predictive capability.
format Article
id doaj-art-7d08e912745842189d8a0d87ab2e0a7d
institution OA Journals
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-7d08e912745842189d8a0d87ab2e0a7d2025-08-20T01:52:03ZengIEEEIEEE Access2169-35362025-01-0113574555746410.1109/ACCESS.2024.340970910549525Software Defect Prediction Based on Effective Fusion of Multiple FeaturesChaozheng Zhang0https://orcid.org/0009-0005-8076-6894Junhua Wu1Nanjing Tech University, Nanjing, ChinaNanjing Tech University, Nanjing, ChinaIn the field of software engineering, the goal of software defect prediction is to accelerate the detection of potential defective modules in products, thereby enhancing the efficiency of allocating testing resources effectively. Traditional software defect prediction models typically rely on a single feature or perform simple aggregation of multiple features during training. However, the former overlooks the potential contributions of other features in defect prediction, while the latter fails to fully exploit the value of different features in the model. Therefore, this paper proposes a software defect prediction model based on the effective fusion of multiple features, named DP-SSCT (Semantics, Source Code, and Tradition). The model integrates traditional features, source code features, and semantic features. To determine the importance of each feature in the model, a gate fusion mechanism and attention mechanism are introduced to capture significant information in the fused features. The proposed model is evaluated on nine open-source Java project samples from the PROMISE dataset. Experimental results demonstrate that the model outperforms existing prediction models, exhibiting superior performance and predictive capability.https://ieeexplore.ieee.org/document/10549525/Software defect predictionconvolutional neural networklong short-term memory (LSTM)bidirectional long short-term memory
spellingShingle Chaozheng Zhang
Junhua Wu
Software Defect Prediction Based on Effective Fusion of Multiple Features
IEEE Access
Software defect prediction
convolutional neural network
long short-term memory (LSTM)
bidirectional long short-term memory
title Software Defect Prediction Based on Effective Fusion of Multiple Features
title_full Software Defect Prediction Based on Effective Fusion of Multiple Features
title_fullStr Software Defect Prediction Based on Effective Fusion of Multiple Features
title_full_unstemmed Software Defect Prediction Based on Effective Fusion of Multiple Features
title_short Software Defect Prediction Based on Effective Fusion of Multiple Features
title_sort software defect prediction based on effective fusion of multiple features
topic Software defect prediction
convolutional neural network
long short-term memory (LSTM)
bidirectional long short-term memory
url https://ieeexplore.ieee.org/document/10549525/
work_keys_str_mv AT chaozhengzhang softwaredefectpredictionbasedoneffectivefusionofmultiplefeatures
AT junhuawu softwaredefectpredictionbasedoneffectivefusionofmultiplefeatures