EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

МОДЕЛЬ РОЗРОБКИ ТЕМАТИЧНИХ ЧАТ-БОТІВ З ВИКОРИСТАННЯМ ПІДХОДУ RAG

Illia Ivasenko, Tetiana Filimonchuk, Stanislav Partyka, Daria Pyvovarova

Системи управління, навігації та зв’язку. Збірник наукових праць · 04.05.2026 · 2 (84) · 110-119

DOI: 10.26906/sunz.2026.2.110 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Системи управління, навігації та зв’язку. Збірник наукових праць, online erschienen am 04.05.2026. Verfasst von Illia Ivasenko und 3 weiteren Autor:innen. Die Arbeit nennt 7 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The relevance of the study is due to the rapid development of large language models (LLM), such as GPT-4, Llama 3 and Claude, which have revolutionized the field of natural language processing (NLP). These models demonstrate exceptional abilities to generate coherent text, generalize information and conduct dialogue. However, when applying LLM in specialized domains (law, medicine, technical support, corporate knowledge bases), critical limitations arise. First, the parametric knowledge of the models is limited by the date of completion of their training (knowledge cutoff), which makes it impossible to work with current information. Second, models are prone to "hallucinations" - the generation of plausible, but actually false statements, especially when the query concerns highly specialized data that is not in the training set. The RAG architecture has become the de facto standard for solving these problems, combining the generative capabilities of LLM with accurate search in external knowledge bases. However, practice shows that the "naive" implementation of RAG (Vanilla RAG) is often insufficient for building reliable systems. Loss of context during search, inability to process complex user queries and lack of verification mechanisms lead to irrelevant answers. In this regard, the current scientific and practical task is not just the implementation of RAG, but the development and research of methods for optimizing each stage of generation. The object of research is the processes of information search and natural language generation in intelligent dialog systems built on the basis of large language models. The subject of the study is methods and algorithms for increasing the accuracy, relevance and contextual consistency of responses in RAG architecture systems, namely hybrid search, improving user queries. Conclusions. The practical value of the study lies in creating an architecture model that can be adapted for any subject area (from technical documentation to regulatory frameworks), while ensuring higher metric accuracy of responses compared to basic solutions.

Zitieren

APA

Ivasenko, I., Filimonchuk, T., Partyka, S., & Pyvovarova, D. (2026). МОДЕЛЬ РОЗРОБКИ ТЕМАТИЧНИХ ЧАТ-БОТІВ З ВИКОРИСТАННЯМ ПІДХОДУ RAG. Системи управління, навігації та зв’язку. Збірник наукових праць, 2(84), 110-119. https://doi.org/10.26906/sunz.2026.2.110

Vancouver

Ivasenko I, Filimonchuk T, Partyka S, Pyvovarova D. МОДЕЛЬ РОЗРОБКИ ТЕМАТИЧНИХ ЧАТ-БОТІВ З ВИКОРИСТАННЯМ ПІДХОДУ RAG. Системи управління, навігації та зв’язку. Збірник наукових праць. 2026;2(84):110-119. doi:10.26906/sunz.2026.2.110

BibTeX
@article{Ivasenko2026,
  author = {Ivasenko, Illia and Filimonchuk, Tetiana and Partyka, Stanislav and Pyvovarova, Daria},
  title = {МОДЕЛЬ РОЗРОБКИ ТЕМАТИЧНИХ ЧАТ-БОТІВ З ВИКОРИСТАННЯМ ПІДХОДУ RAG},
  journal = {Системи управління, навігації та зв’язку. Збірник наукових праць},
  year = {2026},
  volume = {2},
  number = {84},
  pages = {110--119},
  doi = {10.26906/sunz.2026.2.110}
}
Kostenlos starten →