Year: 2025
Type: article
Abstract: With increasingly complex and larger systems, automated testing is nowadays an embedded element of modern development prac-tices. However, root cause analysis (RCA) of test failures remains a time - consuming, knowledge - intensive, and error - prone ac-tivity that typically requires considerable domain expertise. This paper explores the use of an AI - powered RCA on test failure data extracted from Allure reports a standard test framework to generate rich, structured test outcomes. The approach utilizes machine learning methods, natural language understanding, and causality to enable automated diagnosis and classification of failures through the analysis of logs, error messages, and execution traces. It not only accelerates fault localization but also improves the accuracy and uniformity of diagnosis. The review synthesizes current research trends and technology enabling intelligent RCA, re-views current frameworks and tools, and documents the incorporation of AI models into CI/CD pipelines. The paper also addresses significant issues such as interpretability of the model, data quality, and managing flaky tests. By bridging the gap between test result visualization and automated diagnosis, AI - based RCA with Allure presents a smart and scalable solution for improving software reliability and development efficiency.
Author: Alex Thomas Thomas
Language: English
Subfield: Software
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4411501970