Exception Handling Service for Software Agent Ensembles

摘要

The objective of this project has been to develop technology that enables robust operation in open multi-agent systems (MAS) where we have, in principle, no guarantees of reliable, compliant, or even friendly agents and infrastructures, since we have no control or even necessarily any knowledge over how they are implemented. As part of this project we have developed a substantial web-accessible knowledge base of widely usable exception handling expertise, as well as technology that enables (I) the creation of better contingent contracts between agents (both human and software- based) as well as (2) exception-handling agents that monitor MAS for problem symptoms, diagnose the underlying problems, and intervene as appropriate to avoid or resolve these problems.