We attempt to present and compare two methods for representing cases in constraint satisfaction networks. The tension between top-down and bottom-up strategy as regards construction of models of reasoning has long tradition in the literature on AI. The same distinction applies to representations of legal argumentation. Our proposal compares the application of top-down and bottom-up strategy to representation of a chosen judgment of the Court of Justice of the European Union in constraint satisfaction networks. The first (top-down) approach is inspired by systems of defeasible logic and argumentation frameworks. The second approach makes use of bottom-up strategy and is more firmly grounded in the text of the court's decision. The two methods lead to very similar results and yield interesting conclusions concerning the structure of the CJEU reasoning.