Abstract
Large-scale context-aware pervasive environments are dealing with context information of every user for every domain supported. They have to represent every user's personal situation that typically consists of overlapping contexts while caring for the privacy of their users. The contexts are not necessarily all the same for different users that are part of a specific situation. Furthermore, they are perceived differently on base of other contexts that users are, were, or will be involved in; they depend on the personal background and goals of the user. We introduce a generic approach for a social context model and architecture grounded in classic context-aware computing models for context and a cognitive context model approach from discourse analysis. The approach encompasses the users, has distinct categories of context, supports hierarchical context interdependencies, and provides context dependent privacy.