Knowledge-based Software Test Generation

Enriching test oracles with a test expert’s mental model of error prone aspects of software, and granting control to them to specify custom coverage criteria and arbitrary test cases, can potentially improve the quality of automatically generated test suites. This paper reports our investigation on the application of knowledge engineering techniques in automated software testing to increase the the control of test experts on test generation; ontologies and rules are used to specify what needs to be tested and reasoning is used for identification of test objectives, for which test cases are generated. An architecture of the ontology-based approach to testing is presented and a prototype which is implemented for unit testing is described with a case study.