Fault Localization during System Testing

Functional testing of business applications in the enterprise is carried out by independent test teams. Test scripts are generated manually or automatically from requirements, treating the IT systems as a black box. For every release, when test scripts fail to execute, the test teams need to ascertain the cause of failure, which could be due to mismatch between the requirements and the test models and test scripts, or faults in the test scripts or faults in the source code. The process is cumbersome and time consuming. While several techniques have been developed to localize source code faults, these target testing carried out by the developer. To help test teams localize faults, we propose the novel idea of applying source code based fault localization technique to process models that represent the system functionality. Experimental results show that the techniques when applied to models, were able to localize both test script and source code faults.