README.md

metadata

license: cc-by-nc-nd-4.0
language:
  - az
task_categories:
  - text-generation
  - fill-mask
tags:
  - azerbaijan
  - text
  - corpus
  - news
  - books
  - legislation
  - wikipedia
  - localdoc
size_categories:
  - 1M<n<10M

AzTC — Full Version (Azerbaijan Text Corpus)

The expanded version of LocalDoc/AzTC, and one of the largest text corpora in the Azerbaijani language.

Overview

The corpus contains approximately 2.4 billion tokens of Azerbaijani text, compiled and cleaned from a wide range of sources including news portals, books, Wikipedia, and legislation. Text is organized at the document / passage level so that each row is a coherent unit rather than an isolated fragment.

Format

Provided as parquet with the following fields:

• text — the document or passage
• source — provenance label indicating which collection the text came from

License

The dataset is licensed under CC BY-NC-ND 4.0:

• Attribution — you must give appropriate credit and indicate if changes were made.
• Non-Commercial — you may not use the material for commercial purposes.
• No Derivatives — if you remix, transform, or build upon the material, you may not distribute the modified material.

See the CC BY-NC-ND 4.0 license for details.

Contact

For questions or issues, please contact LocalDoc at v.resad.89@gmail.com

Citation

@misc{aztc_full,
  title  = {Azerbaijan Text Corpus — Full Version (AzTC)},
  author = {LocalDoc},
  howpublished = {\url{https://huggingface.co/datasets/LocalDoc/AzTC}},
  note   = {Licensed under CC BY-NC-ND 4.0}
}