Agent-based systems co-operation

Information exchange between sites within network is currently based on automated data exchange at the protocol level and unfortunately only on supervised information exchange at the user level. Automation of information exchange is as difficult, as still no domain-independent general semantics can be defined users of small systems, as well as those of large interactive systems that are connected to the network, suffer from insufficient support for their information exchange. In order to overcome this semantics problem, we propose a framework in which users can incrementally define and store their knowledge about distributed data and information sources. For this purpose, we have developed the AgentTeam framework where this provides for network connectivity at the information level. It is designed to support the co-operative work of user teams, users, and network sites in a distributed and heterogeneous environment. AgentTeam simulates the user a distributed but homogeneous database environment. In this contribution, we introduce the architecture of AgentTeam and a sample prototype implementation. We discuss the prototype by showing how users can build their own views by composing their data from autonomous locally dispersed DBMSs.