Overview of technologies used in intelligent dialog systems

Abstract

The purpose of the research is to review existing intelligent systems for providing customer service. Voice assistants for customer support over the phone and text chat bots are being considered. The research method was the collection of information from open Russian and foreign sources. Results. A review of existing machine learning technologies used in dialog systems designed to serve customers in various areas of commercial activity was carried out. The methods on which these technologies are based are given. The areas of activity in which such software products are used are described. The article discusses the problems of the subject area, as well as the prospects for the development of intelligent dialogue systems. The article presents the results of psychological research on the topic of interaction between a robot and a person, as well as security problems when using intelligent systems, such as leakage of personal data or information constituting a trade secret. One of the areas is protecting users from telephone spam. Conclusion. During the work, it was revealed that approaches to the implementation of dialogue systems can be different: systems with a limited set of ready-made answers; hybrid systems «machine-human»; systems based on deep learning that can hold the context of a dialogue; systems focused on extracting motivating words from speech or text. With all the variety of methods and techniques, all dialogue systems have common features. A diagram of the decomposition of the dialogue system is presented. The article will be useful to developers of voice dialogue systems and text chat bots in which human-machine interaction occurs.