Schema Evolution publication categorizer

One critical foundation for reliable collective coordinated action in multi-agent systems is the ability to exchange representational objects that can be locally interpreted in ways that make collective sense. Coherent local interpretation of shared representations is needed to create global semantic coherence for the distributed actions of individual agents. The process of autonomously establishing a collective semantics and organization of the concepts in an open domain is not well understood. This paper discusses our motivations and approach to this problem. It presents a specific formalization and algorithm that we call mutual online ontology alignment with its rationale, and illustrates the performance of the approach with some simulation experiments. Keywords: ontology evolution, ontology alignment, colearning