Scholarlink

<p>The reliability and testability of the software could be enhanced by effectively predicting faults in the early phases of software development. The inherent complexity of service-oriented architectures and the frequent communication between services, generates large amount of logs and fault data. Several studies have investigated machine learning methods to predict faults in service-oriented architecture based systems. However, they often overlook how source code metrics can be effectively used to predict faults. This study introduces an efficient algorithm called the Adaptive Fractional Water Cycle Algorithm (Adaptive Fr-WCA) that integrates Fractional Calculus, the Water Cycle Algorithm, and an Adaptive mechanism to enhance fault prediction performance. The proposed methodology begins by transforming Web Services Description Language files into Java files. Faults are then injected into these Java files to create a dataset for training. A Deep-Maxout Network (DMN) is subsequently trained using the Adaptive Fr-WCA to perform fault prediction. Experimental results using real-world web service data demonstrate that the proposed Adaptive Fr-WCA-based DMN model significantly outperforms traditional approaches. The proposed method achieved a recall of 91.90%, a F-measure of 92.10%, and a precision of 92.0%, highlighting its effectiveness and robustness in fault prediction tasks.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Software Fault Prediction in Service-Oriented Architecture Using Deep Learning Enabled Adaptive Optimization Technique

  • 期刊论文

  • 发表日期：

    2026年2月6日

查看全文
  • Raghuraj Singh,
  • Kuldeep Kumar

摘要

The reliability and testability of the software could be enhanced by effectively predicting faults in the early phases of software development. The inherent complexity of service-oriented architectures and the frequent communication between services, generates large amount of logs and fault data. Several studies have investigated machine learning methods to predict faults in service-oriented architecture based systems. However, they often overlook how source code metrics can be effectively used to predict faults. This study introduces an efficient algorithm called the Adaptive Fractional Water Cycle Algorithm (Adaptive Fr-WCA) that integrates Fractional Calculus, the Water Cycle Algorithm, and an Adaptive mechanism to enhance fault prediction performance. The proposed methodology begins by transforming Web Services Description Language files into Java files. Faults are then injected into these Java files to create a dataset for training. A Deep-Maxout Network (DMN) is subsequently trained using the Adaptive Fr-WCA to perform fault prediction. Experimental results using real-world web service data demonstrate that the proposed Adaptive Fr-WCA-based DMN model significantly outperforms traditional approaches. The proposed method achieved a recall of 91.90%, a F-measure of 92.10%, and a precision of 92.0%, highlighting its effectiveness and robustness in fault prediction tasks.