Understanding and maintaining the structure of large ontologies is a cognitively demanding task. Visualizations are commonly used as a cognitive aid for presenting large ontologies and instance data. One challenge with these visual techniques is that the generated views are often very dense and complex. It is difficult for users to navigate to concepts within the overall structure that meet the needs of the current task. In this paper, we present our work towards understanding the cognitive support requirements for navigating ontologies and our application of these requirements to the design of a plugin for the popular ontology editor Prot´eg´e.