Regression Test cases selection using Natural Language Processing

• Published in: 2020 International Conference on Intelligent Engineering and Management (ICIEM) pp. 301 - 305
• Main Authors: Sutar, Shantanu, Kumar, Rajesh, Pai, Sriram, BR, Shwetha
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-06-2020
• Subjects: Artificial intelligence; Computer bugs; History; Natural language processing; Organizations; Regression testing; Software testing; Test efficiency; Test prioritization; Text similarity
• DOI: 10.1109/ICIEM48762.2020.9160225

Regression Testing is one of the important phases to detect the effects of new development or modifications done in the already existing product. As the product grows, the number of regression test cases also increases to manifold. In an agile world, it is very important to extract test cases which are having very high potential to find defects to reduce the overall release cycle. In practice, there are many ways to select test cases based on different criteria. Many of them are based on historical defects in the product as historical defect clusters can be one of defect prone areas because of defect fixes. However, considering the high number of historical defects it becomes difficult to select test cases merely based on defect clusters or any other static techniques. In this paper, we propose our approach to find the high potential regression test cases from the master test suite using Natural Language Processing by selecting a test case based on its intent match with defects. The application developed from this solution has helped us in reducing the regression cycle and enhanced the exploratory productivity for our product. This method also opens the door for new concepts like generating test cases automatically based on its learnings from the product's historical defects, existing test cases, and new feature development.