Scholarlink

In both human-centric and software-driven projects, ensuring the stability of systems after changes is essential. Regression testing verifies that recent modifications don’t negatively affect existing functionalities. This mirrors human processes, where workflow or policy changes must be tested for smooth operations. Test Case Selection (TCS) is critical in optimizing regression testing by identifying and prioritizing relevant test cases, reducing redundancy. Integrating Machine Learning (ML) algorithms can enhance TCS, selecting cases most likely to detect faults, saving time and resources, much like streamlining human decision-making. This research evaluates two ML clustering techniques for TCS: Execution-Spectra-Based Sampling (ESBS) and Weighted Attribute-Based Strategy (WAS), focusing on their human-centric implications in selecting event-sequence-based test cases. The research follows a five-phase framework: literature review, problem formulation, data collection, examination, and conclusion. Our findings show that the WAS technique significantly outperforms ESBS in identifying issues in event-sequence test cases. Specifically, WAS achieved a 100% failure detection ratio for two programs, Circular Queue (CQ) and Bank, while ESBS achieved 76.92% and 57.14%, respectively. These results indicate that WAS is more effective at prioritizing test cases with higher failure detection, improving the efficiency and accuracy of event-sequence selection. Centering the role of human testers in using these ML techniques, this research aims to boost both efficiency and decision-making in software testing teams. The goal is to empower teams to integrate advanced ML techniques while preserving the human insight needed for effective software quality assurance.

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

Evaluating Human-Centric Machine Learning Clustering Techniques for Selecting Event Sequence Test Cases

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nik Ilhami Nik Ibrisam,
  • Nadiah Mohd Hanim,
  • Johanna Ahmad,
  • Eg Su Goh

摘要

In both human-centric and software-driven projects, ensuring the stability of systems after changes is essential. Regression testing verifies that recent modifications don’t negatively affect existing functionalities. This mirrors human processes, where workflow or policy changes must be tested for smooth operations. Test Case Selection (TCS) is critical in optimizing regression testing by identifying and prioritizing relevant test cases, reducing redundancy. Integrating Machine Learning (ML) algorithms can enhance TCS, selecting cases most likely to detect faults, saving time and resources, much like streamlining human decision-making. This research evaluates two ML clustering techniques for TCS: Execution-Spectra-Based Sampling (ESBS) and Weighted Attribute-Based Strategy (WAS), focusing on their human-centric implications in selecting event-sequence-based test cases. The research follows a five-phase framework: literature review, problem formulation, data collection, examination, and conclusion. Our findings show that the WAS technique significantly outperforms ESBS in identifying issues in event-sequence test cases. Specifically, WAS achieved a 100% failure detection ratio for two programs, Circular Queue (CQ) and Bank, while ESBS achieved 76.92% and 57.14%, respectively. These results indicate that WAS is more effective at prioritizing test cases with higher failure detection, improving the efficiency and accuracy of event-sequence selection. Centering the role of human testers in using these ML techniques, this research aims to boost both efficiency and decision-making in software testing teams. The goal is to empower teams to integrate advanced ML techniques while preserving the human insight needed for effective software quality assurance.