Abstract (EN)

Recent research on multi-party communications shows how multi-agent communications can take advantage of the complexity of the human communication process. The salient point is the very nature of the communication channels which enable humans to focus their attention on ambient communications, as well as to direct their own communications. For multi-agent systems, the difficulty is the routing of messages according to both the needs of the sender and the needs of the (potential) recipients . This difficulty is compounded by the necessity of taking into account the context of this communication. This article proposes an architecture for the Environment as Active Support for Interaction model (EASI) which is based on a classification data model and supports multiparty communication. Our proposition has been implemented and the functional description of the environment is given.