An approach to embedding of agents into knowledge representation and processing system

Abstract

The paper presents an approach to the development of a software environment for creation of intelligent systems. This environment is based on an integrated knowledge representation model that combines both classic knowledge representation means and constraint programming methods with agent-based techniques. In contrast to other systems following constraint programming methods, the environment presented here allows one to manipulate objects with imprecise (subdefinite) values of attributes. The process of refining such subdefinite attributes is performed by means of a special data-driven mechanism. Another feature of the environment is the use of an agent-based technique instead of the traditional production rule technique, in order to increase efficiency of the processes of logical inference and data processing. Due to a natural combination of the data-driven and event-driven mechanisms, the environment can be used for the development of efficient intelli-gent systems for various applications.