A Model-Driven Approach to Formalize and Support Controlled Experiments in Software Engineering

In this paper, we present the main goals, questions and hypotheses of our research related to provide automated support environments to conduct controlled experiments in software engineering (SE). Here, we introduce a model-driven approach that has been developed for supporting the formalization and execution of SE experiments. It consists of: (i) a domain specific language (DSL) for process specification and statistical design of controlled experiments; (ii) model-driven transformations that promote the generation of workflow models specific to each experiment participant and according to the experiment statistical design; and (iii) a workflow execution environment that allows the monitoring of participant activities in the experiment. In addition, we present the planning of empirical studies for this approach. The research was divided in phases and each phase has empirical studies defined in order to validate it and answer our research questions. Each phase adopts cycles of successive approximations (generateand-test cycles).