What question did this study set out to answer?

The research aims to develop and analyze a corpus for aspect extraction in scientific texts from multiple domains including Russian and Kazakh.

SciMDIX: A dataset for aspect extraction from multi-domain scientific documents in Kazakh and Russian

View Full Paper

Ask AI

Bookmark

Authors

This research unveils a dataset for aspect extraction in scientific texts, suggesting improvements for low-resource languages.

The research aims to develop and analyze a corpus for aspect extraction in scientific texts from multiple domains including Russian and Kazakh.
Created a dataset of 412 abstracts annotated for aspect-based information extraction in Kazakh and Russian.
Analyzed aspect distribution across four scientific domains: IT, linguistics, medicine, and psychology.
Conducted experiments using deep learning, mBERT, and XLM-RoBERTa + CRF architecture.
Achieved effective zero-shot aspect extraction between Russian and Kazakh using multilingual models.
Annotated datasets include 2,129 aspects in Russian and 2,027 aspects in Kazakh across seven categories.

Shvarts et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark