A Review of Software Testing Process Log Parsing and Mining

The substantial log data present in software testing processes are often the primary and sometimes the sole source of information for test engineers to understand and debug software. Consequently, analysis and mining of log data play a crucial role, especially in the application of these techniques. Testers can utilize this data to identify scenarios or anomalies not covered in tests, thereby improving test cases, enhancing test coverage and effectiveness, and ultimately bolstering software safety and reliability. However, extracting valuable information from this abundant data poses a challenge, as logs generated during testing processes are often unstructured and not easily computable. To address this challenge, researchers are actively engaging in deep mining and analysis of test data for digital processing. In this article, we provide a comprehensive review of existing literature, systematically categorizing and detailing log mining and analysis in software testing. This includes methods for converting logs into structured event templates and using logs for anomaly detection, failure prediction, and diagnostic assistance. Integrating current research challenges and future directions, we discuss and forecast the prospects of research in this field.