Combining behavior and data modeling in automated test case generation

Abstract

Software testing plays a critical role in the process of creating and delivering high-quality software products. Manual software testing can be an expensive, tedious and error-prone process, therefore testing is often automated in an attempt to reduce its cost and improve its defect detection capability. Model-based testing, a technique used in automated test case generation, is an important topic because it addresses the need for test suites that are of high-quality and yet, maintainable. Current model-based techniques often use a single model to represent system behavior. Using a single model may restrict the number and type of test cases that may be generated. In this paper, system-level test case generation is accomplished using two models to represent system behavior. The results of case studies used to evaluate this technique indicate that for the systems studied a larger percentage of the required test cases can be generated using the combined modeling approach.