This study implements the deep bi-affine neural dependency parsing for the Uzbek language using a available Universal Dependency treebank. Key aspects such as attention layers, network architecture, embedding dropout rates, and optimization parameters (e.g., Adam optimizer) are discussed comprehensively. Experimental results demonstrate the bi-affine model achieves a UAS of 79.5% and a LAS of 72.4%, significantly outperforming a transition-based baseline parser by 6.7% (UAS) and 7.4% (LAS).
| Mualliflar | Matlatipov, Sanatbek |
|---|---|
| Jurnal | Al-Farg'oniy avlodlari |
| Nashr sanasi | 2025-03-23 |
| Son | 1 |
| Betlar | 143-148 |
| Til | Rus |
синтаксический анализ зависимостей, би-аффинное внимание, Узбекский язык, низкоресурсные языки, нейронные сети, Universal Dependencies, BiLSTM, dependency parsing, Uzbek language, bi-affine attention, BiLSTM, neural networks, low-resource, NLP (Natural Language Processing), Universal Dependencies, O'zbek tili, bi-affin neyron modeli, BiLSTM, chuqur neyron tarmoqlari, kam resursli tillar universal bog‘liqliklar
E-mail systems remain one of the primary means of communication among users. However, unwanted messages pose a serious threat to system security and represent a significant risk to users and organizations. This paper…
This article analyzes the method of hacking biometric recognition systems through print attacks, and also proposes an algorithm for detecting fake facial images based on the analysis of color textures. For this purpose…
This study investigates the hydrothermal treatment of wheat grain to optimize moistening conditions and enhance flour quality. Utilizing both conventional analytical methods and advanced computational modeling…
The article studies the business cycle model for e-commerce companies in the development of the e-commerce sector, the changes in its stages under the influence of IT, the stages of the commercial cycle of companies in…
The article focuses on developing methods for preventing cyberattacks in cloud environments, analyzing modern approaches to cloud security and mathematical models. Cloud technologies play a crucial role in ensuring the…
Abstract: The article presents a comparative analysis of the digitalization process using the example of the world's leading libraries, including the Library of Congress of the United States, the British Library, the…
. This study investigates the thermo-mechanical behavior of a functionally graded material (FGM) cylindrical rod subjected to torsional loading and temperature gradients. The problem is solved using a perturbation…
This article provides an overview of automated clinic systems and their main types. The importance of automation in healthcare lies in improving service quality, streamlining processes, and saving time. The article…
The aim of this study is to develop a comprehensive list of syntactic and morphological tags in Uzbek, which aims to create a dataset for natural language processing (NLP) problems in Uzbek. Based on existing tagset…
This article provides an in-depth analysis of advanced algorithms for automatic object detection and classification in video imagery. The rapid advancements in artificial intelligence and deep learning have…