Enterprise Architecture Management has been proposed to help organizations in their efforts to flexibly adapt to rapidly changing market environments. Enterprise architectures are described by means of conceptual models depicting, e.g., an enterprise?s business processes, its organisational structure, or the data the enterprise needs to manage. Such models are stored in large repositories. Using these repositories to support enterprise transformation processes often requires detecting structural patterns containing particular labels within the model graphs. As an example, consider the case of mergers and acquisitions. Respective patterns could represent specific model fragments that occur frequently within the process models of the merging companies. This paper introduces an approach to analyse conceptual models at a structural and semantic level. In terms of structure, the approach is able to detect patterns within the model graphs. In terms of semantics, the approach is able to detect previously standardized model labels. Its core contribution to enterprise architecture management and transformation is two-fold. First, it is able to analyse conceptual models created in arbitrary modelling languages. Second, it supports a wide variety of pattern-based analysis tasks related to managing change in organisations. The approach is applied in a merger and acquisition scenario to demonstrate its applicability.