When it comes to the judgment of expertise, we often rely on assumptions which propose that expert knowledge is structured differently than novice knowledge. However, regardless of how much descriptive plausibility there might be in single cases, hypotheses like this cannot be investigated across domains unless there is a way to subtract the content from the structure which allows testing for structure only. A new algorithm introduced in this chapter is designed to solve this problem. The algorithm works for small and medium graphs, is capable of completely mapping the structure of an undirected graph, and also allows one to compare full structure sets between pairs of graphs. In my discussion of this algorithm, I derive additional computing-efficient graph feature-based heuristics from the original algorithm and compare the two in order to show how large graphs can be analyzed in a similar way. I then present standard applications of the algorithm from empirical studies on different kinds of expertise. Finally, I provide examples and a guideline in which the application and the interpretation of the structure comparison measure are discussed with a focus on research practice.