Agent-Based Software Engineering and Agent Mediations

Abstract

Techniques are presented for designing with software agents and new concepts entitled Abstract Intelligent Maintenance and implementations. The designs present concurrent agent algebras with diagnosis learning, faults, and maintenance capability. The stages of conceptualization, design and implementation are defined with AI agents and Mediators, Multiagent implementations are proposed to support software design techniques, which incorporates object level nondeterministic knowledge learning and knowledge representation developed in the projects. Objects, message passing actions, and implementing agents are defined by syntactic constructs, with agents appearing as functions. Agent-based maintenance fault recovery, maintenance events and activity are computed for the AII agents. The techniques provide a basis for automatic implementation with agents. Mediators implementing objects and agents define Interportability. The object level definitions for individual modules are turned to executable programs applying source abstract syntax tree to target abstract syntax tree morphisms. The techniques are applied to define a Preservation Principle for Agent-based Maintenance and heterogeneous KB basis to learning diagnosis.