Dataset · Zenodo · October 18, 2026
UzPoetMeta Gold v1.1: Open Annotations for Metaphor Detection in Classical and Modern Uzbek Poetry
Dauletov Adilbek Yusupbayevich, Noila Matyakubova
Abstract
UzPoetMeta Gold v1.1: Open Annotations for Metaphor Detection in Classical and Modern Uzbek Poetry UzPoetMeta Gold v1.1 is an expert-approved annotation dataset developed for computational research on metaphor detection and figurative language analysis in classical and modern Uzbek poetry. The dataset contains 1,500 annotated poetic units, including 750 units from Alisher Navoiy's Hayrat ul-abror and 750 units from Eshqobil Shukur's Hamal ayvoni. The gold-standard annotations comprise 911 METAPHORICAL, 546 LITERAL, and 43 AMBIGUOUS labels. The release includes expert-approved annotations, semantic categories, source metadata, annotation decisions, historical benchmark partitions, model predictions, and evaluation results. A historical binary benchmark of 1,193 units is provided with 835 training, 176 development, and 182 test records. Expert auditing identified same-poem overlaps between historical partitions. Therefore, the included benchmark results should not be interpreted as performance on a fully poem-disjoint independent test set. Data availability and copyright: This public release does not contain original poetic texts or copyrighted literary excerpts. The original annotation labels, curated metadata, and derived evaluation materials are distributed under Creative Commons Attribution 4.0 International (CC BY 4.0), insofar as the dataset creators hold the relevant rights. Third-party literary works and specific editions or transcriptions are excluded from this license. Citation: Users should provide appropriate attribution to the dataset creators and cite the Zenodo DOI in academic publications. Citation of the related research article is also requested when available.
More by Noila Matyakubova
- A Multitask XLM-R-Based Hybrid Approach for Named Entity Recognition, Coreference Resolution, and Genealogical Relation Extraction from Uzbek Historical DocumentsMachine Learning and Knowledge Extraction · 2026
- UzArchMeta-IR: A Metadata-Aware Retrieval Test Collection for Uzbek Digital ArchivesZenodo · 2026
- UzAlign v1.1.1: A Morphologically Annotated Uzbek–English Parallel Corpus with Model-Assisted Word AlignmentData · 2026
- ArchiveGene-RealZenodo · 2026
