Abstract: This article describes a multi-agent software architecture to manage
a corporate memory in the form of a corporate semantic Web. It summarizes the
design rationale and the final architecture of the CoMMA system. It then
presents and discusses an approach to manage distributed annotations, focusing
on two problems: the allocation of newly posted annotations and the allocation
of the tasks involved in distributed query-solving.
Keywords: multi-agent information systems, corporate memory, semantic Web, ontology