Две модели вопросно-ответной «машины диалога»

• Date: 2020
• Main Authors: Верлань, Анатолій Федорович, Чмир, Ігор Олексійович
• Format: Article
• Language: Ukrainian
• Published: Kamianets-Podilskyi National Ivan Ohiienko University 2020
• Online Access: http://mcm-tech.kpnu.edu.ua/article/view/216422
• Journal Title: Mathematical and computer modelling. Series: Technical sciences

The article describes two models of question-answering dialogue. Both models describe the behavior of an active agent (an agent, who is asking questions) of a question-answering dialogue in the process of solving a problem by means of a dialogue process. In this sense, the models are intended to describe the script of a dialogue method for solving a problem. It is assumed that with the help of dialogue methods it is possible to solve problems, which belong to the class of ill-formalized. The first model is based on the idea of the Mealy state machine. It has a limited applicability, since when generating the next question, the history of previous answers of the reactive agent is not taken into account. The second model is based on the idea of a Petri net. The model takes into account the history of previous answers of the reactive agent, as well as the possibility of non-standard processing of the question-answering transaction using processes-daemons. The models are problem-independent and model a question-answering dialogue, which does not depend on the subject area of the dialogue and the nature of the problems solved by means of the dialogue method. The problem-independence of the models is due to the fact that the standard, cyclically repeated actions for processing a question-answering transaction reflect a unified cognitive dialogue cycle. The unified cognitive dialogue cycle is built on the basis of Neisser's cyclical model of perception. In the cognitive dialogue cycle, in contrast to the Neisser’s cycle of perception, the source of the flow of sensory events is not the external environment, but the reactive agent. The models are designed to specify «dialogue machine» software that simulates the purposeful behavior of an active dialogue agent in the process of solving a problem. Both models assume that the main components of the «dialogue machine» are not mapped into a program code, but represented by data stored in a database. A dialogue machine is thus a software system that implements a script of a dialogue method by interpreting the data stored in a database.