Agent-based Service-Oriented Intelligent P2P Networks for Distributed Classification

The paper addresses a new class of distributed intelligent systems emerging at the cross boundary of a number of new advanced topics of modern information technology, i.e. peer-to-peer systems, multi-agent systems, distributed data mining and classification, service-oriented architectures, etc. The paper objective is, based on proof-of-concept case studies, to emulate and evaluate the service search and delivery processes using distributed "yellow pages" of agents operating on peer-to-peer basis, while assuming that the services are specified in FIPA-like format. The paper is developing an idea of using agent-based service oriented intelligent networks for distributed classification. It also presents two software prototypes of the "proof-of-concept" level explicating the problem peculiarities and demonstrating the proposed solutions.