Syntactic tagging of sentences based on integrated neural networks
Boburkhon TuraevDepartment of Computer Systems, Tashkent University of Information Technologies named after Muhammad Al-Khwarizmi, Tashkent, UzbekistanBahodir ZaripovDepartment of Computer Systems, Tashkent State University of Economics, Tashkent, UzbekistanMaxsud MaxammadiyevDepartment of Computer Systems, Tashkent State University of Economics, Tashkent, UzbekistanGavhar TurayevaDepartment of Economics, Tashkent International University of Education, Tashkent, UzbekistanDilnoza ShabonovaDepartment of Economics, Tashkent State Transport University, Tashkent, UzbekistanXurshid TurayevDepartment of Computer Systems, Tashkent University of Information Technologies named after Muhammad Al-Khwarizmi, Tashkent, Uzbekistan
2025
ABI
Abstract
This research presents the development of an integrated RNN–CYK model designed for the syntactic analysis of Uzbek sentences based on neural networks. The proposed RNN–CYK model employs a two-stage approach: first, the RNN architecture learns the dynamic context of sentences, and then the CYK algorithm constructs a syntactic tree based on strict grammatical rules. During model training, the processes of POS tagging, text normalization, tokenization, lemmatization, and embedding were applied sequentially. Experimental results show that the developed system achieved accuracy – 0.85, BLEU – 0.87, ROUGE-L – 0.79, METEOR – 0.93, and ChrF++ – 0.91.
Topics
Identifiers
Citations and references
Cited by 011 references