A Linguistic Feature-Based Model and Algorithm for Assessing Author Writing Style in Uzbek Texts
Abstract
This paper presents a linguistic feature-based model and algorithm for analyzing author writing style in Uzbek texts. The approach is motivated by the agglutinative structure of Uzbek, in which one lexical stem can generate many surface word forms through derivational and inflectional suffixes. Conventional Type-Token Ratio (TTR), when calculated only over surface forms, may therefore overestimate lexical diversity. To address this, the proposed pipeline calculates lexical richness at token, stem, and lemma levels, extracting morphology-aware features such as inflectional variety, derivational variety, POS distribution, and suffix usage. In response to the need for practical validation, the paper includes a clear processing schema, formal pseudocode, a corpus-based case study on five thematic Uzbek texts, and a baseline comparison with traditional TTR. Finally, the proposed architecture is implemented as a Django-based prototype featuring automated feature extraction, segmentation, and visual reporting.