Cerberose/vietnamese-classification-dataset

Vietnamese Classification Dataset

The Vietnamese Classification Dataset is a large-scale corpus of Vietnamese text samples annotated with categorical labels across three distinct classes. It was created to support research and development in Vietnamese Natural Language Processing (NLP), particularly in text classification, topic modeling, and supervised learning tasks. With over 11 million examples, this dataset offers the scale and linguistic diversity necessary for training robust, data-driven language models and classification systems for Vietnamese.

This dataset builds upon our crawled data from Vietnamese-English news sites and publicly available resources including the Vietnamese News Corpus, Vietnamese Correction v2, and OPUS Tatoeba. We thank the respective authors and maintainers for their contributions to open Vietnamese NLP research.

If this dataset contributes to your research or applications, please include the following citation:

@misc{cerberose_vietnamese_classification_dataset_2025,
title = {Vietnamese Classification Dataset},
author = {Cerberose},
year = {2025},
howpublished = {\url{https://huggingface.co/datasets/Cerberose/vietnamese-classification-dataset}}
,
note = {Accessed: YYYY-MM-DD}
}

Source Datasets

This dataset incorporates and extends data from the following public sources:

@misc{binhvq_news_corpus_2020,
  title        = {Vietnamese News Corpus},
  author       = {Binh VQ},
  year         = {2020},
  howpublished = {\url{https://github.com/binhvq/news-corpus}},
  note         = {Accessed: 2024-12-22}
}

@misc{bmd1905_vietnamese_correction_v2_2023,
  title        = {Vietnamese Correction v2},
  author       = {BMD1905},
  year         = {2023},
  howpublished = {\url{https://huggingface.co/bmd1905/vietnamese-correction-v2}},
  note         = {Accessed: 2024-12-23}
}

@misc{wecover_opus_tatoeba_2021,
  title        = {OPUS Tatoeba Parallel Corpus},
  author       = {WeCover},
  year         = {2021},
  howpublished = {\url{https://huggingface.co/datasets/wecover/OPUS_Tatoeba}},
  note         = {Accessed: 2024-12-24}
}

This dataset is distributed under the Apache 2.0 License. Users should review and comply with the license terms before applying the data to academic, commercial, or derivative works.

For questions, feedback, or contributions, please contact the maintainer Cerberose or open an issue on the dataset’s Hugging Face page. Contributions such as additional documentation, class clarification, or dataset refinement are welcome and encouraged.

Dataset URL: https://huggingface.co/datasets/Cerberose/vietnamese-classification-dataset