Abstract : Workflows are increasingly adopted to describe large-scale data- and compute-intensive processes that can take advantage of today's Distributed Computing Infrastructures. Still, most Scientific Workflow formalisms are notoriously difficult to fully exploit, as they entangle the description of scientific processes and their implementation, blurring the lines between what is done and how it is done as well as between what is and what is not infrastructure-dependent. This work addresses the problem of data-intensive Scientific Workflow design by describing scientific experiments at a higher level of abstraction, emphasizing scientific concepts over technicalities, easing the separation of functional and non-functional concerns and leveraging domain knowledge. To achieve this goal, we propose a model-driven approach enhanced with Knowledge Engineering technologies. The main contributions of this work are a semantic Scientific Workflow model to capture user goals and a generative process assisting the transformation from high-level models to executable workflow artefacts.