metadata

license:
  - apache-2.0
annotations_creators:
  - no-annotation
language_creators:
  - found
language:
  - kk
multilinguality:
  - monolingual
source_datasets:
  - original
task_categories:
  - text-generation
  - fill-mask
pretty_name: MKD | Multi-Domain Kazakh Dataset

Dataset Description

Point of Contact: Sanzhar Murzakhmetov, Besultan Sagyndyk

Dataset Summary

MKD | Multi-Domain Kazakh Dataset is a Kazakh-language dataset containing just over 21 911 503 unique texts from multiple domains.

Supported Tasks

'MLM/CLM': can be used to train a model for casual and masked languange modeling

Languages

The kk code for Kazakh as generally spoken in the Kazakhstan

Data Instances

For each instance, there is a string for the text and a string for the id.

{'id': '0054d6d30dbcad772e20b22771153a2a9cbeaf62',
 'text': '«Челсидің» бапкері Жозе Моуриньо қызметінен кетті Лондондықтар Жозе Муриньюмен қош айтысты. Себеп көп. Англия чемпиондары маусымды өте сүреңсіз бастады. Ұпайды оңды солды шашқан ақсүйектер турнир кестесінің төменгі сатысында қалып қойды.'

Data Fields

id: a string which is hexidecimal hash for text in split
text: a string containing the content body

Data Splits

The Multi-Domain Kazakh dataset has 5 splits: cc100-monolingual-crawled-data, kazakhBooks, leipzig, oscar and kazakhNews. Below are the statistics of the dataset:

Dataset Split	Domain	Number of texts in Split	Number of tokens in Split	Number of unique tokens in Split	Median number of tokens in text
cc100-monolingual-crawled-data	Wikipedia articles	19 569 201	440 737 547	6 071 847	12
kazakhBooks	Books	4 481	179 627 716	4 649 820	38 443
leipzig	Articles/News	1 734 397	27 013 206	1 115 876	14
oscar	CommonCrawl	269 965	229 713 848	3 781 885	427
kazakhNews	News	333 459	87 606 307	1 255 834	177

With overall stats:

Stat	Value
Number of texts	21 911 503
Number of tokens	966 795 903
Number of unique tokens	11 674 839

Full dataset takes 21GB

Annotations

The dataset does not contain any additional annotations.

Personal and Sensitive Information

Datset is not anonymized, so individuals' names can be found in the dataset. Information about the original author is not included in the dataset.

Social Impact of Dataset

The purpose of this dataset is to organize open-source datasets on the Kazakh language for further research and commercial uses

Licensing Information

The Multi-Domain Kazakh dataset version 1.0.0 is released under the Apache-2.0 License.

Datasets:

kz-transformers
/

multidomain-kazakh-dataset