Autonomous classification via self-formation of collections in AuInSys

Abstract

This paper describes a mechanism, called SIF, for autonomous and unsupervised hierarchical classification of information items in AuInSys, a pro-interactive information system, where each single information element is represented by a compact software agent. The classification is carried out through the self-organization of information items which form the collections incrementally. The information items compete with each other through the self-interactions of their agents which lead to incremental grouping of items into collections. The proposed mechanism provides an automatic, more flexible, and dynamic partitioning of information compared with what is used in traditional file systems. © Springer Science+Business Media B.V. 2008.