Morphological analysis of the karakalpak language using neural networks
Abstract
This article explores methods for morphological analysis of the Karakalpak language using neural networks. Karakalpak, a Turkic language with agglutinative morphology, presents challenges for traditional rule-based approaches. We propose architectures based on recurrent (RNN) and transformer (Transformer) networks for tasks such as lemmatization, grammatical category identification, and morpheme segmentation. Quantitative results are presented using the [dataset name] dataset, along with comparisons to classical methods (Finite-State Morphology)
About the Authors
List of references
Камалов С. Каракалпакская морфология: традиционный анализ. — 2020.
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. — 2019.
Google’s Tensor2Tensor for Seq2Seq. — 2018.
Утеулиев Н., Кудайбергенов Ж. Распознавание речевых эмоций на основе дополняющих сетей и Wav2vec 2.0. — Ташкентский университет информационных технологий имени Мухаммада ал-Хоразмий, Нукусский филиал, 2024.
Khudaybergenov, K., & Bakhritdinov, F. Physics-Informed Neural Network with Multidimensional Weight Connections for Differential Equations // Raqamli Transformatsiya va Sun’iy Intellekt Ilmiy Jurnali. — 2024. — Vol. 2, Issue 3, June.
URL: https://dtai.tsue.uz/index.php/dtai/article/view/v2i32/v2i32
Утеулиев Н., Кудайбергенов Ж. Обучение языковых моделей RNN на основе гипотез автоматического распознавания речи // Amaliy matematikaning zamonaviy muammolari va istiqbollari Respublika ilmiy-amaliy konferensiya materiallari. — Qarshi Davlat Universiteti, 24–25 май 2024. — 152 b.