Conceptual Graphs (CG), originally developed for static data representation have been extended to cope with dynamical aspects. This paper adresses two questions connected with the topic: How can implicational knowledge about a system’s states and behaviour be derived from a dynamic CG description and how can the CG specification process be supported by automatic or semiautomatic algorithms? Based on Formal Concept Analysis (FCA) we propose methods for both problems. Guided by an example we introduce two kinds of formal contexts containig the dynamic system’s information: state contexts and action contexts. From these the complete implicational knowledge can be derived. Combining the techniques of attribute exploration and determination of a formal context’s concepts, we demonstrate a procedure which interactively asks for the validity of implications and from this information designs a dynamic CG system with the desired properties.