Chimir I. Modeling of dialogue agents in intellectual systems

This dissertation is devoted to a theoretical generalization and solving of a scientific problem - modeling and synthesis of dialogue agents oriented toward solving ill-formalized problems, - which is important for the theory and practice of systems and tools of artificial intelligence. A cognitive model of erotetic dialogue process based on fundamental Neisser's cy-cle of perception is proposed. On the basis of this model a series of models of a dialogue method of access to the memory of questions of an active agent of the dialogue are elabo-rated. Logical and datalogical models of question-answering pairs are proposed and on their basis a model of a Dialogue Knowledge Base (DiKB) is generated. The deductive properties of the DiKB and the equivalence of the processes of logical inference and ero-tetic dialogue are investigated. The possibility of the DiKB to store the procedural knowl-edge of a human-expert, and the adaptability of an erotetic dialogue to solve ill-formalized problems in various systems of applied artificial intelligence are also investigated. An ar-chitecture of a Dialogue Problem Solver is proposed.