Building an Open Multi-Author Uzbek Literary Corpus: A Reusable Resource for NLP and Literary Research
Year: 2026
Type: conference-paper
Abstract: Open and well-structured literary corpora are essential for advancing natural language processing in low-resource languages, yet such resources remain scarce for Uzbek. This paper describes the development of an open multi-author Uzbek literary corpus intended for computational linguistics, digital humanities, and language technology research. The corpus comprises approximately 700 literary works written by 20 prominent authors, totaling more than 1.84 million tokens. To ensure consistency and computational usability, the texts were collected, normalized, encoded in UTF-8, and enriched with standardized metadata through a unified preprocessing workflow. A hierarchical organization at the author and document levels enables efficient retrieval, corpus exploration, and statistical analysis. Beyond supporting corpus-based linguistic and stylometric studies, the dataset provides a valuable literary resource for educational purposes and future domain adaptation of Uzbek language models. This paper details the corpus architecture, construction process, preprocessing procedures, and descriptive characteristics, with the aim of facilitating transparent and reproducible research. The corpus is publicly available through the Uzbek Poetic Corpus platform at https://uzbekpoetikcorpus.uz/.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7224919656