Abstract
Whatever purpose models are created for, they need to be evaluated in order to ensure that they will serve their intended use. Unfortunately, model evaluation is a significant effort and no standard clear-cut evaluation methodology exists today. In this work we introduce our approach of distributing the evaluation effort over the whole modeling process. This is achieved by providing (partly automated) feedback on the models in different stages of development. We illustrate our approach based on the specic goal of modeling tasks, topics and skills for the adaptive work-integrated learning system APOSDLE.