metadata

annotations_creators:
  - no-annotation
language:
  - crh
license: cc-by-4.0
multilinguality:
  - monolingual
task_categories:
  - text-generation
  - fill-mask
task_ids:
  - language-modeling
  - masked-language-modeling
pretty_name: CrimeanTatarMonocorpus
tags:
  - crh
  - books
  - corpus

Crimean Tatar corpus

Overview

In an effort to democratize research on low-resource languages, we release CrimeanTatarMonocorpus dataset, a books corpus consisting of materials from 100+ unique sources in the Crimean Tatar Language. All annotation was done for Crimean Tatar corpus project. The text is provided as is, so some pre-processing is needed. For example, some subtitles contain timestamps information. Both Cyrillic and Latin alphabets are used in text, so to switch between them you could use this transliterator.

Dataset Name: Crimean Tatar corpus
Language: Crimean Tatar
Domain: General (books, subtitles, etc.)

To load and use dataset, run this script:

from datasets import load_dataset

crh_corpus=load_dataset("QIRIM/crh_monocorpus")