Designing a Domain-Specific Semantic Network for Uzbek Mining Terminology: A Computational Lexicography Approach

This study presents the development of "Konchilik terminlarining zamonaviy lug‘ati" (Modern Dictionary of Uzbek Mining Terms), a semantically structured and ontology-driven thesaurus designed to standardize and digitize mining terminology in the Uzbek language. Informed by frame semantics, domain expert validation, and NLP-compatible modeling, the project addresses key challenges such as semantic ambiguity, synonym redundancy, and multilingual interference—issues particularly prevalent in post-Soviet technical vocabularies. Over 3,000 terms were analyzed and hierarchically structured using Protégé, with relationships formalized through classes, subclasses, and object properties reflecting semantic roles. To validate the practical usability of the resource, a semantic retrieval prototype was developed using FAISS indexing and the RAG (Retrieval-Augmented Generation) architecture. This prototype enables semantic search in Uzbek mining texts by embedding user queries and matching them against vectorized terminology. The resource thus supports machine-readable dictionary construction, semantic role labeling, and intelligent retrieval, contributing to national NLP infrastructure and serving as a scalable model for domain-specific terminology engineering in under-resourced languages.

Перевод пока недоступен