Work

Achieving higher accuracy in classifying Uzbek words into grammatical categories using the CRF model

Year: 2025

Type: conference-paper

Abstract: The classification of words into grammatical categories (part-of-speech tagging) is a fundamental task in natural language processing (NLP). For morphologically rich languages such as Uzbek, this process becomes more challenging due to complex affixation, agglutinative word forms, and limited resources. This paper investigates the application of the Conditional Random Fields (CRF) model for Uzbek word classification, aiming to achieve higher accuracy compared to traditional approaches such as Hidden Markov Models (HMM). By incorporating contextual and morphological features, CRF achieves more reliable tagging. Several Uzbek sentence examples are analyzed, with CRF applied step by step to demonstrate its advantages. Experimental results show that CRF significantly improves accuracy, achieving 92.7% compared to 84.3% for HMM.

DOI: 10.70413/si-v03-i34-s34292

Authors: Sami Kobilov, Javohir Nazarov, Ilyos Rabbimov

Institution: Samarkand State University named after Sharof Rashidov

Language: English

Topic: Natural Language Processing Techniques

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

OpenAlex ID: W7083205039

FWCI:

Cites:

Cited by:

Related to: