EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

TOOLS FOR SEMANTIC SEARCH AND ANSWER GENERATION IN UKRAINIAN-LANGUAGE MUSEUM SYSTEMS

Khrystyna LIPIANINA-HONCHARENKO, Vadym VITENKO, Diana ZAHORODNIA

Computer systems and information technologies · 31.05.2026 · (2) · 70-77

DOI: 10.31891/csit-2026-2-7 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer systems and information technologies, online erschienen am 31.05.2026. Verfasst von Khrystyna LIPIANINA-HONCHARENKO und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This article examines modern semantic search models and a search-based response generation approach for building intelligent museum information systems focused on Ukrainian-language and bilingual textual materials. It is shown that traditional lexical search, based on keyword matching, does not always ensure adequate quality in cultural heritage tasks, where relevance is determined not only by formal term matching but also by the semantic proximity of the query and the document. In this context, text vector representation models take on particular significance, as they allow texts to be represented in vector space and enable the retrieval of contextually relevant information even in the absence of direct lexical matches. This paper analyzes the LaBSE, paraphrase-multilingual-MiniLM-L12-v2, multilingual-E5-large-instruct, and BGE-M3 models in terms of Ukrainian language support, multilingualism, search accuracy, computational resource requirements, and suitability for local deployment. The generation of search-based responses is examined separately as an approach in which a generative model constructs a response based on fragments retrieved by a search module from an external knowledge base. It is argued that this approach is particularly suitable for museum information systems, as it reduces the risk of hallucinations, improves the factual accuracy of responses, and enables a source verification mechanism. The advantages and limitations of open and closed generative models in the context of Ukrainian-language museum services are also systematized. Based on the analysis, practical recommendations are formulated regarding the selection of search models and the configuration of answer generation systems for digital cultural heritage.

Zitieren

APA

LIPIANINA-HONCHARENKO, K., VITENKO, V., & ZAHORODNIA, D. (2026). TOOLS FOR SEMANTIC SEARCH AND ANSWER GENERATION IN UKRAINIAN-LANGUAGE MUSEUM SYSTEMS. Computer systems and information technologies, 70-77. https://doi.org/10.31891/csit-2026-2-7

Vancouver

LIPIANINA-HONCHARENKO K, VITENKO V, ZAHORODNIA D. TOOLS FOR SEMANTIC SEARCH AND ANSWER GENERATION IN UKRAINIAN-LANGUAGE MUSEUM SYSTEMS. Computer systems and information technologies. 2026:70-77. doi:10.31891/csit-2026-2-7

BibTeX
@article{LIPIANINAHONCHARENKO2026,
  author = {LIPIANINA-HONCHARENKO, Khrystyna and VITENKO, Vadym and ZAHORODNIA, Diana},
  title = {TOOLS FOR SEMANTIC SEARCH AND ANSWER GENERATION IN UKRAINIAN-LANGUAGE MUSEUM SYSTEMS},
  journal = {Computer systems and information technologies},
  year = {2026},
  number = {2},
  pages = {70--77},
  doi = {10.31891/csit-2026-2-7}
}
Kostenlos starten →