PRINCIPLES OF DEVELOPMENT OF PARALLEL CORPUS-BASED LINGUISTIC RESOURCES FOR UZBEK AND CHINESE LANGUAGES

The article substantiates the principles of developing parallel corpus-based linguistic resources for the Uzbek–Chinese language pair, which remains one of the least resourced combinations in Central Asian computational linguistics despite the rapid growth of political, economic and educational ties between Uzbekistan and China. On the basis of a comparative analysis of recent Uzbek, Chinese and multilingual corpus projects, the study identifies the typological, technological and organizational factors that determine the quality of a bilingual resource: the agglutinative structure and two-script tradition of Uzbek, the absence of word boundaries and the classifier system in Chinese, the scarcity of authentic translations, and the risks of synthetic data. Seven principles are proposed, namely representativeness and balance, authenticity and directionality, multi-level alignment, interoperable annotation, quality assurance, legal and ethical transparency, and openness and sustainability, together with a six-stage development model that integrates human expertise with embedding-based alignment tools.