<![CDATA[Current computational processing of Uzbek language is delayed by the limitations of "black-box" statistical architectures in handling high-density agglutinative morphology. While neural models excel in analytic languages, they face a definitive "Out-of-Vocabulary (OOV) glass ceiling" when confronted with the generative capacity of Uzbek suffixation. This paper introduces a high-precision, rule-based morphological analyzer built upon Uzbek-Lex, a gold-standard lexicon of 83,646 entries mapped to Universal Dependencies (UD) v2 standards. Our system explicitly models linguistic "stress points," including 7,370 verbal nouns and 777 multi-category roots, utilizing deterministic syntactic constraints for disambiguation. Empirical benchmarking against the Stanza Uzbek model demonstrates a reduction in the OOV rate from 15.0% to 5.8%. These results suggest that for morphologically rich, low-resource languages, deterministic linguistic verification is a scientific necessity for high-fidelity dependency parsing.]]>
Copyrights © 2026
