Digital access to large amounts of scholarly text presents both challenges and opportu nities for researchers in the humanities.Meeting these challenges depends on having high-quality representations of the contents of digital resources suitable for both machines and humans to use.Different ways of categorizing these contents are appropriate for different purposes, leading to the further problem of relating the contents of different categorization schemes to each other.This essay discusses the rationale for categorizing philosophical concepts and surveys some of the main approaches to doing so for materials that are continuously changing.It describes the goals and methods of the Indiana Philos ophy Ontology (InPhO) project and provides an example of the kind of analysis that is made possible by powerful modeling methods.