metadata

language:
  - kk
license: cc-by-4.0
multilinguality:
  - monolingual
size_categories:
  - 100K<n<1M
task_categories:
  - token-classification
task_ids:
  - named-entity-recognition
paperswithcode_id: kaznerd
pretty_name: 'KazNERD: Kazakh Named Entity Recognition Dataset'
viewer: true
dataset_info:
  config_name: KazNERD
  features:
    - name: index
      dtype: string
    - name: sentence_id
      dtype: string
    - name: tokens
      sequence: string
    - name: ner_tags
      sequence:
        class_label:
          names:
            '0': O
            '1': B-ADAGE
            '2': I-ADAGE
            '3': B-ART
            '4': I-ART
            '5': B-CARDINAL
            '6': I-CARDINAL
            '7': B-CONTACT
            '8': I-CONTACT
            '9': B-DATE
            '10': I-DATE
            '11': B-DISEASE
            '12': I-DISEASE
            '13': B-EVENT
            '14': I-EVENT
            '15': B-FACILITY
            '16': I-FACILITY
            '17': B-GPE
            '18': I-GPE
            '19': B-LANGUAGE
            '20': I-LANGUAGE
            '21': B-LAW
            '22': I-LAW
            '23': B-LOCATION
            '24': I-LOCATION
            '25': B-MISCELLANEOUS
            '26': I-MISCELLANEOUS
            '27': B-MONEY
            '28': I-MONEY
            '29': B-NON_HUMAN
            '30': I-NON_HUMAN
            '31': B-NORP
            '32': I-NORP
            '33': B-ORDINAL
            '34': I-ORDINAL
            '35': B-ORGANISATION
            '36': I-ORGANISATION
            '37': B-PERSON
            '38': I-PERSON
            '39': B-PERCENTAGE
            '40': I-PERCENTAGE
            '41': B-POSITION
            '42': I-POSITION
            '43': B-PRODUCT
            '44': I-PRODUCT
            '45': B-PROJECT
            '46': I-PROJECT
            '47': B-QUANTITY
            '48': I-QUANTITY
            '49': B-TIME
            '50': I-TIME
  splits:
    - name: train
      num_bytes: 17067134
      num_examples: 90228
    - name: validation
      num_bytes: 2113552
      num_examples: 11167
    - name: test
      num_bytes: 2126698
      num_examples: 11307
  download_size: 9016377
  dataset_size: 21307384
configs:
  - config_name: KazNERD
    data_files:
      - split: train
        path: train.arrow
      - split: validation
        path: valid.arrow
      - split: test
        path: test.arrow

KazNERD: Kazakh Named Entity Recognition Dataset

KazNERD is a corpus of 112,702 sentences extracted from television news text, annotated with 136,333 named entities across 25 entity classes. Annotation was performed by two native Kazakh speakers using the IOB2 scheme. The data are provided in the CoNLL 2002 format.

Dataset Description

Homepage: https://issai.nu.edu.kz/kaznerd-eng/
Repository: https://github.com/IS2AI/KazNERD
Paper: https://aclanthology.org/2022.lrec-1.44
Point of Contact: Rustem Yeshpanov

Models

Models with following architecures were trained on the dataset: CRF, BiLSTM-CNN-CRF, BERT, and XLM-RoBERTa. The best-performing XLM-RoBERTa model achieved an exact match F1-score of 97.22% on the test set.

Citation

@inproceedings{yeshpanov-etal-2022-kaznerd,
    title = "{K}az{NERD}: {K}azakh Named Entity Recognition Dataset",
    author = "Yeshpanov, Rustem  and
      Khassanov, Yerbolat  and
      Varol, Huseyin Atakan",
    booktitle = "Proceedings of the Thirteenth Language Resources and Evaluation Conference",
    month = jun,
    year = "2022",
    address = "Marseille, France",
    publisher = "European Language Resources Association",
    url = "https://aclanthology.org/2022.lrec-1.44",
    pages = "417--426",
    abstract = "We present the development of a dataset for Kazakh named entity recognition. The dataset was built as there is a clear need for publicly available annotated corpora in Kazakh, as well as annotation guidelines containing straightforward{---}but rigorous{---}rules and examples. The dataset annotation, based on the IOB2 scheme, was carried out on television news text by two native Kazakh speakers under the supervision of the first author. The resulting dataset contains 112,702 sentences and 136,333 annotations for 25 entity classes. State-of-the-art machine learning models to automatise Kazakh named entity recognition were also built, with the best-performing model achieving an exact match F1-score of 97.22{\%} on the test set. The annotated dataset, guidelines, and codes used to train the models are freely available for download under the CC BY 4.0 licence from https://github.com/IS2AI/KazNERD.",
}

https://github.com/IS2AI/KazNERD