Methods and Algorithms of POS-tagging of Adverbs and Pronouns in Uzbek Texts

In this scientific article, the problem of automatic identification of adverbs and pronouns in Uzbek texts is analyzed based on linguistic and computational approaches. For the Uzbek language, which has an agglutinative structure, the task of POS-tagging causes not only morphological, but also syntactic difficulties. In particular, the multifunctional and contextual variability of adverbs and pronouns requires precise approaches to their automatic classification. The article analyzes performance indicators between traditional rule-based algorithms, as well as modern models based on statistical and neural networks (Conditional Random Fields and BiLSTM). Methodologically, the formal classification of models, their functionality and experimental foundations. Based on the results, the advantages and weaknesses of each approach are identified, and proposals for the optimal tagging model for the Uzbek language are put forward. The research results will be of great importance for morphological analysis, machine translation, automatic text indexation, and improving the overall quality of Uzbek NLP systems.

Перевод пока недоступен