Datasets:

Kencorpus
/

KenCorpus_text

language:
  - sw
  - luo
  - bxk
  - lri
  - rag
license: cc-by-4.0
task_categories:
  - text-generation
  - text-classification
tags:
  - swahili
  - kiswahili
  - dholuo
  - lubukusu
  - lumarachi
  - lulogooli
  - kenyan-languages
  - low-resource-languages
  - african-languages
  - text-corpus
pretty_name: KenCorpus Text
size_categories:
  - 1K<n<10K

KenCorpus Text: A Kenyan Multilingual Text Corpus

Dataset Description

KenCorpus Text is a multilingual text corpus for Kenyan languages, collected from language communities including indigenous stories, student compositions, native language media stations, and publishers. The corpus goes beyond conventional religious texts to represent everyday language use.

Three languages were selected: Kiswahili, Luhya (dialects: Lumarachi, Logooli, Lubukusu), and Dholuo.

Dataset Statistics

Language	Files	Source
Swahili	2,585	Community texts
Swahili (Tweets)	324	Social media
Dholuo	166	Community texts
Lumarachi	449	Community texts
Lubukusu	16	Community texts
Logooli	247	Community texts

Data Fields

Column	Type	Description
text	string	Full text content
language	string	Language (Swahili, Dholuo, Lubukusu, Lumarachi, Logooli, Luhya_Bukusu, etc.)
genre	string	Genre (Culture, News, Creative_writing, Exam, Agriculture, etc.)
source	string	Data source (Community, Educational_Institution, Media, Publisher, etc.)
title	string	Title of the text if available
story_id	string	Unique identifier matching the KenCorpus metadata
source_type	string	Source category (community_text, tweets)

Example Record

{
    'text': 'HISTORIA YA SHULE YA UPILI YA OLEFREMA...',
    'language': 'Swahili',
    'genre': 'Creative_writing',
    'source': 'Educational_Institution',
    'title': 'Historia ya shule',
    'story_id': '3510',
    'source_type': 'community_text'
}

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("Kencorpus/KenCorpus_text")

# Access samples
print(dataset['train'][0])

# Filter by language
swahili = dataset['train'].filter(lambda x: x['language'] == 'Swahili')
dholuo = dataset['train'].filter(lambda x: x['language'] == 'Dholuo')

print(f"Swahili: {len(swahili)}, Dholuo: {len(dholuo)}")

Genres

Culture, News, Creative_writing, Exam, Agriculture, Commerce, Religion, Song, Health, Story, and more.

Data Collection

Primary data was collected from respective language communities, including indigenous stories and narratives from student compositions, native language media stations, and publishers.

Dataset Curators

Kiswahili: Rose Felynix, Khalid Kitito, Dr. Benard Okal Dholuo: Jotham Ondu Ajiki, Dr. Jackline Okello, Jonathan Muga, Mercy Lavinca Oduoll Luhya (Logooli): Salano Odari, Dr. Phillip Lumwamu Luhya (Bukusu): Mactilda Nekesa Makana, Mulwale Martin Luhya (Marachi): Yonah Weunda

Citation

@article{wanjawa2022kencorpus,
  title={Kencorpus: A Kenyan Language Corpus of Swahili, Dholuo and Luhya for Natural Language Processing Tasks},
  author={Wanjawa, Barack W. and Wanzare, Lilian D. and Indede, Florence and McOnyango, Owen and Ombui, Edward and Muchemi, Lawrence},
  journal={arXiv preprint arXiv:2208.12081},
  year={2022}
}

License

CC-BY-4.0

Acknowledgments

Part of the Kencorpus project for low-resource Kenyan language NLP resources.