This chapter provides an overview of approaches relating to the development of qualitative models of regulatory networks, via approaches similar to model-checking and to logic programming. It focuses on studying this problem in a large-scale context, in other words, potentially with several hundred interacting components. The chapter discusses logical approaches for learning about dynamic biological systems. It then focuses on inductive logic programming methods making it possible, from time series data, to build a logical model of a control network. The chapter also discusses the various acceptable hypotheses in terms of logical learning, including non-determinism. It presents an approach, based on the Answer Set Programming paradigm, allowing the inference regulation networks with delays from temporal data. The chapter ends by recognizing the benefits and limits of the various existing logical learning methods and draws up research perspectives for the years to come.