|
|
--- |
|
|
license: apache-2.0 |
|
|
task_categories: |
|
|
- text-generation |
|
|
tags: |
|
|
- legal |
|
|
size_categories: |
|
|
- 10M<n<100M |
|
|
--- |
|
|
# Swahili Text Dataset |
|
|
|
|
|
## Overview |
|
|
|
|
|
This dataset contains a comprehensive collection of Swahili text data, derived from the [AfriBERTa Corpus](https://huggingface.co/datasets/castorini/afriberta-corpus). It provides a rich resource for natural language processing tasks focused on the Swahili language. |
|
|
|
|
|
## Dataset Details |
|
|
|
|
|
- **Source**: [AfriBERTa Corpus](https://huggingface.co/datasets/castorini/afriberta-corpus) (Swahili subset) |
|
|
- **Language**: Swahili |
|
|
- **Size**: [Insert total number of samples here] |
|
|
- **Format**: Hugging Face Dataset |
|
|
|
|
|
## Content |
|
|
|
|
|
The dataset consists of two main columns: |
|
|
|
|
|
1. `id`: A unique identifier for each text entry |
|
|
2. `text`: The Swahili text content |
|
|
|
|
|
## Usage |
|
|
|
|
|
You can load this dataset using the Hugging Face `datasets` library: |
|
|
|
|
|
```python |
|
|
from datasets import load_dataset |
|
|
|
|
|
dataset = load_dataset("[Your_HuggingFace_Username]/[Your_Dataset_Name]") |
|
|
``` |
|
|
|
|
|
Replace `[Your_HuggingFace_Username]` and `[Your_Dataset_Name]` with the appropriate values for your uploaded dataset. |
|
|
|
|
|
## Data Fields |
|
|
|
|
|
- `id`: string |
|
|
- `text`: string |
|
|
|
|
|
## Data Splits |
|
|
|
|
|
This dataset combines training and test splits from the original AfriBERTa Corpus. The data has been shuffled with a fixed seed (42) to ensure reproducibility. |
|
|
|
|
|
## Dataset Creation |
|
|
|
|
|
This dataset was created by: |
|
|
|
|
|
1. Loading the Swahili subset of the AfriBERTa Corpus |
|
|
2. Concatenating the training and test splits |
|
|
3. Shuffling the combined dataset |
|
|
4. Extracting the 'id' and 'text' fields |
|
|
|
|
|
## Intended Uses |
|
|
|
|
|
This dataset can be used for various natural language processing tasks involving the Swahili language, such as: |
|
|
|
|
|
- Language modeling |
|
|
- Text classification |
|
|
- Named entity recognition |
|
|
- Machine translation (as a source or target language) |
|
|
- Sentiment analysis |
|
|
- And more... |
|
|
|
|
|
## Limitations |
|
|
|
|
|
- The dataset is limited to the content available in the original AfriBERTa Corpus. |
|
|
- It may not represent all dialects or variations of the Swahili language. |
|
|
- The quality and accuracy of the text content depend on the original data source. |
|
|
|
|
|
## Citation |
|
|
|
|
|
If you use this dataset, please cite the original AfriBERTa Corpus: |
|
|
|
|
|
``` |
|
|
@inproceedings{ogueji-etal-2021-small, |
|
|
title = "Small Data? No Problem! Exploring the Viability of Pretrained Multilingual Language Models for Low-resourced Languages", |
|
|
author = "Ogueji, Kelechi and |
|
|
Zhu, Yuxin and |
|
|
Lin, Jimmy", |
|
|
booktitle = "Proceedings of the 1st Workshop on Multilingual Representation Learning", |
|
|
month = nov, |
|
|
year = "2021", |
|
|
address = "Punta Cana, Dominican Republic", |
|
|
publisher = "Association for Computational Linguistics", |
|
|
url = "https://aclanthology.org/2021.mrl-1.11", |
|
|
pages = "116--126", |
|
|
} |
|
|
``` |
|
|
|
|
|
## Licensing Information |
|
|
|
|
|
This dataset is derived from the AfriBERTa Corpus. For usage terms and conditions, please refer to the [original dataset's license](https://huggingface.co/datasets/castorini/afriberta-corpus). |
|
|
|
|
|
## Contact |
|
|
|
|
|
If you have questions or comments about this specific version of the dataset, please open an issue in this repository or contact [[email protected]]. |
|
|
|
|
|
--- |
|
|
|
|
|
Dataset created and curated by [AdeptSchneider]. |
|
|
Last updated: [09/10/2024] |
