Enterprise information systems must be populated with large-scale live data to ensure the testing and evaluating effectiveness, but current development processes typically implement test-data generation as an isolated task and incorporate very little automation for this step. The paper proposes a novel framework that seamlessly integrates data analysis, simulation, and automated generation into the life cycle of object- oriented software development. Along with system analysis, design and implementation, statistical properties of objects are identified, modeled, and built into the software products; afterward simulation and test tools can automatically distill these properties, generate accurate simulation data, and perform system tests close to their real-world usage. The paper also presents SDGen, an implementation of an automated software tool for the framework.
展开▼