An Information Retrieval-Based Regression Test Selection Technique

Abstract

Regression testing (RT) is the crucial part of the software testing process. It is applied after a bug fix or a change in the functionality of the codebase. The main goal is to ensure that the modified software has the desired outcome and does not cause adverse effects in other parts of the software. RT may be costly depending on the test’s quantity and complexity. Therefore, regression test selection (RTS) can be introduced to minimize these costs. RTS runs only the test cases related to the modified parts of the software. Currently, various RTS studies focus on compiled languages such as Java, C/C++, and C#, and they mostly rely on direct code dependency between tests and the system under test. In this study, we have introduced a new RTS tool called Smartest to reduce the number of selected integration tests. Former RTS tools were focused mainly on unit tests according to dependencies of modified source files. Smartest is the first RTS tool that works for software written in JavaScript and can select integration tests written in natural language by the quality assurance team. Smartest is tested on three commercial projects and observed that it picks 13% of all test cases on average. Experiments show that Smartest minimizes the selected integration tests on RTS processes, although it does not use file-level code dependency. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023.