A Computational Framework for Text Complexity Analysis in Uzbek Secondary Education
Year: 2026
Type: conference-paper
Abstract: This paper examines the problem of assessing the difficulty of Uzbek texts for students in grades 9–11. The relevance of this topic lies in the fact that in school practice, texts are not always selected based on their actual comprehension difficulty. The same text may be understandable in length but complex in its wording, sentence structure, or hidden cultural meanings. Therefore, this paper proposes a computational approach that allows for assessing text difficulty based on several factors simultaneously. The proposed method includes three main criteria, the first of which involves identifying complex words and phrases using a dictionary-based approach. The second criterion is based on syntactic analysis of the text using Stanza and allows for the identification of complex sentence structures. The third criterion focuses on identifying allusions using the mBERT model. A separate percentage is then calculated for each criterion, and these percentages are used to determine the overall difficulty level of the text. To test the proposed approach, Uzbek texts of various genres were used: educational, popular science, journalistic, fiction, historical and cultural, and socio-analytical. The results showed that the system handles different types of text differently. Educational and popular science materials proved the easiest to analyze, while fiction and historical and cultural texts, which are more likely to contain hidden meanings, complex syntax, and allusions, proved the most challenging.
Language: English
Subfield: Linguistics and Language
Field: Social Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: hybrid
OpenAlex ID: W7202285178