Explanation constraints in a multi-agent environment

• Autores: C. Piquemal-Baluard, P. Glize
• Localización: Panel '92: actas, XVIII Conferencia Latinoamericana de Informática, 1992, págs. 969-979
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • The explicit separation of the knowledge and the system that uses them (the inference's engine) offers to expert systems dialog abilities forbidden to programs built on traditional data processing technics.

    This capacity of dialog and self-explanation must be used and developed because it represents one of the main advantages of this kind of system. Futhermore the qualities of user friendship involved in the installation of a mechanism of explanation, this allows paticularly the systems to be validated. The justification of the results increases in a large way the confidence accorded by user to the solution found.

    We present a study of the explanation in the context of a multi-agent system with one unique explanatory module. Because of the constraints of this kind of software, the explanation cannot be based on the addition of specific explanatory knowledge on each form that can be, but only on syntactic analysis of the knowledge and user model. We analyse the consequences of this approach and we give examples on each part in the medical field.