Software testing is one of the essential and expensive tasks in software development. Hence, many approaches were introduced to automate different software testing tasks. Among these techniques, search-based test generation techniques have been vastly applied in real-world cases and have shown promising results. These strategies apply search-based methods for generating tests according to various test criteria such as line and branch coverage. In this thesis, we introduce new search objectives and techniques using various knowledge carved from resources like source code, hand-written test cases, and execution logs. These novel search objectives and approaches (i) improve the state-of-the-art in search-based crash reproduction, (ii) present a new search-based approach to generate class-integration tests covering interactions between two given classes., and (iii) introduce two new search objectives for covering common/uncommon execution patterns observed during the software production.
|Qualification||Doctor of Philosophy|
|Award date||22 Apr 2021|
|Publication status||Published - 2021|
- Search-based Software Testing
- Crash Reproduction
- Class Integration Testing
- Carving Information Sources