DS4H-ICTU/yat-ner-dataset

Yambeta Named Entity Recognition (NER) Dataset

Dataset Description

This dataset was developed for Named Entity Recognition (NER) tasks in Yambeta, a Bantu language from Cameroon. The dataset contains sentences from the Yambeta Bible text corpus, annotated with named entities such as persons, locations, and organizations.

• —Developed by: DS4H-ICTU Research Group in cooperation with the Yambeta Bible Project.
• —Language(s): Yambeta (Bantu language from Cameroon)
• —License: Apache 2.0 (or specify if different)
• —Dataset Type: Annotated Text (NER)

Dataset Sources

• —Source Text: Yambeta Bible text corpus (final_dataset.xlsx)
• —Number of Samples: {total_samples} samples
• —Named Entities: {totalentities} unique entities across {entitytypes} categories (persons, locations, organizations)

Uses

• —Direct Use: This dataset can be used for training, fine-tuning, and evaluating NER models in Yambeta.
• —Downstream Use: Models trained on this dataset can be utilized for applications like translation, entity recognition, or information extraction.

Bias, Risks, and Limitations

• —Biases: The dataset is extracted from a religious text and may not represent a comprehensive set of Yambeta language entities.
• —Out-of-Scope Use: The dataset may not generalize well to non-religious texts in Yambeta.

Dataset Structure

• —Format: JSON or DatasetDict with tokens and ner_tags fields.
• —Train/Validation/Test Split:
• —Train set: 80%
• —Validation set: 10%
• —Test set: 10%

Citation

If you use this dataset in your work, please cite it using the following format:

@misc{yambeta_ner_dataset,
  title = {Yambeta NER Dataset},
  author = {Dr.-Ing. Philippe Tamla},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/DS4H-ICTU/yat-ner-datasets}
}

Contact Information

For more information, contact the developers at: philiptamla@gmail.com