• 1 Laboratoire LISCO

The existence of crosscutting concerns tangled or scattered, complicates the understanding and evolution of object oriented source code. The industrial adoption of aspect-oriented paradigm has led to research new approaches supporting aspect oriented migration. This migration requires the identification of crosscutting concerns, in order to encapsulate them into aspects. We propose in this paper a new approach for the identification of crosscutting concerns at the conceptual level. We materialize this latter by the UML class and sequence diagrams. We use the formal concept analysis to group scattered functionalities in sequence diagrams, and we analyze the order of method calls to detect the tangled ones. Then, we filter all obtained candidate aspects, in order to avoid the mistakes.