Адаптивний бот-консультант для систем електронної комерції на основі нейромережевої архітектури «ТРАНСФОРМЕР
The article conducts a comprehensive study of the problem of automating customer communication in modern e-commerce systems. It is determined that existing scenario-based approaches to building chatbots have exhausted their potential given the growing demands for personalization and speed of service...
Збережено в:
| Дата: | 2026 |
|---|---|
| Автори: | , |
| Формат: | Стаття |
| Мова: | Українська |
| Опубліковано: |
Vinnytsia National Technical University
2026
|
| Теми: | |
| Онлайн доступ: | https://oeipt.vntu.edu.ua/index.php/oeipt/article/view/866 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | Optoelectronic Information-Power Technologies |
| Завантажити файл: |  |
Репозитарії
Optoelectronic Information-Power Technologies| Резюме: | The article conducts a comprehensive study of the problem of automating customer communication in modern e-commerce systems. It is determined that existing scenario-based approaches to building chatbots have exhausted their potential given the growing demands for personalization and speed of service. A new information technology for creating an intelligent consultant bot is proposed, based on a hybrid combination of Large Language Models (LLM) of the Transformer architecture and RAG (Retrieval-Augmented Generation) methodology. This approach enables natural response generation using up-to-date product data stored in a relational database. The shortcomings of existing commercial solutions are analyzed in detail, and the advantages of the developed system are formulated |
|---|---|
| DOI: | 10.31649/1681-7893-2026-51-1-117-129 |