We are interested in the problem of data sharing in overlay networks that have a social structure, i.e. participants are linked and exchange data with others w.r.t. the similarity of their data semantics. In this paper we propose a methodology to produce conceptual synopses for the semantics that are encapsulated in the schemas of relational data that are shared in a social network. These synopses are constructed solely based on semantics that can be deduced from schemas themselves with some optional additional conceptual clarifications. The produced synopses represent in a concentrated way the current semantics. Existing or new participants can refer to these synopses in order to determine their interest in the network. We present a methodology that employs the conceptual synopsis for the construction of a mediating schema. These can be Used as global interface,; for sharing of information in the social network. Furthermore, we extend our methodology in order to compress the conceputal synopsis such that infrequent concepts are eliminated and the respective inferred global schema encapsulates the most popular semantics of the social network.