Analysis of Models for Building Dependency Parsing in Agglutinative Languages
Abstract
In the domain of computational linguistics, syntactic analysis that is, the comprehension of the interrelationship between words-plays a pivotal role in natural language processing (NLP). Dependency parsing, a fundamental component of computational linguistics, offers a methodical approach to syntactic analysis. On a global scale, one of the most advanced and vital branches of NLP is dependency parsing, which focuses on detecting dependency relations between words. This technique is of particular significance in applications such as machine translation and sentiment analysis. The present article discusses the construction of dependency parsing models for agglutinative languages, their application to Uzbek sentence structures, the differences from other language types, and an analysis of foundational data resources and models for the Uzbek language. In this study, a BERT-based dependency parsing model is employed to analyze Uzbek sentence structures. The article discusses the construction of dependency parsing models for agglutinative languages and their application to the Uzbek language, highlighting differences from other language types. Furthermore, it provides an analysis of foundational linguistic resources and pretrained language models relevant to Uzbek. The paper delineates the distinguishing characteristics of grammatical dependencies in Uzbek sentences by leveraging contextual embeddings generated by the BERT model and juxtaposes these findings with traditional syntactic parsing approaches. Finally, the study evaluates the advantages and future potential of BERT-based dependency parsing for low-resource and agglutinative languages such as Uzbek.