Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
parquet
Sub-tasks:
topic-classification
Size:
10K - 100K
ArXiv:
License:
metadata
annotations_creators:
- expert-annotated
language:
- kin
- run
license: mit
multilinguality: multilingual
source_datasets:
- andreniyongabo/kinnews_kirnews
task_categories:
- text-classification
task_ids:
- topic-classification
dataset_info:
- config_name: kinnews_cleaned
features:
- name: text
dtype: string
- name: label
dtype: int64
splits:
- name: train
num_bytes: 20011123
num_examples: 9075
- name: test
num_bytes: 4432204
num_examples: 2008
download_size: 13396388
dataset_size: 24443327
- config_name: kirnews_cleaned
features:
- name: text
dtype: string
- name: label
dtype: int64
splits:
- name: train
num_bytes: 2828871
num_examples: 1749
- name: test
num_bytes: 87818
num_examples: 65
download_size: 1629277
dataset_size: 2916689
configs:
- config_name: kinnews_cleaned
data_files:
- split: train
path: kinnews_cleaned/train-*
- split: test
path: kinnews_cleaned/test-*
- config_name: kirnews_cleaned
data_files:
- split: train
path: kirnews_cleaned/train-*
- split: test
path: kirnews_cleaned/test-*
tags:
- mteb
- text
Kinyarwanda and Kirundi news classification datasets (KINNEWS and KIRNEWS, respectively), which were both collected from Rwanda and Burundi news websites and newspapers, for low-resource monolingual and cross-lingual multiclass classification tasks. Each news article can be classified into one of 14 possible classes: politics, sport, economy, health, entertainment, history, technology, culture, religion, environment, education, relationship.
| Task category | t2c |
| Domains | News, Written |
| Reference | https://arxiv.org/abs/2010.12174 |
Source datasets:
Dataset Preparation in MTEB
This repository is a staging copy of andreniyongabo/kinnews_kirnews for MTEB. The intended long-term canonical benchmark copy is mteb/KinNewsClassification.
Transformations
- Built
text = title + " " + content - Preserved the two source configs:
kinnews_cleanedandkirnews_cleaned - Applied dataset cleaning before upload to reduce duplicates and train-test leakage in the benchmark copy
Label Schema
- Integer class labels from the source dataset are preserved
Splits and subsets
kinnews_cleaned: cleaned train/test splitskirnews_cleaned: cleaned train/test splits
How to evaluate on this task
You can evaluate an embedding model on this dataset using the following code:
import mteb
task = mteb.get_task("KinNewsClassification")
evaluator = mteb.MTEB([task])
model = mteb.get_model(YOUR_MODEL)
evaluator.run(model)
To learn more about how to run models on mteb task check out the GitHub repository.
Citation
If you use this dataset, please cite the dataset as well as mteb, as this dataset likely includes additional processing as a part of the MMTEB Contribution.
@article{niyongabo2020kinnews,
author = {Niyongabo, Rubungo Andre and Qu, Hong and Kreutzer, Julia and Huang, Li},
journal = {arXiv preprint arXiv:2010.12174},
title = {KINNEWS and KIRNEWS: Benchmarking Cross-Lingual Text Classification for Kinyarwanda and Kirundi},
year = {2020},
}
@article{enevoldsen2025mmtebmassivemultilingualtext,
title={MMTEB: Massive Multilingual Text Embedding Benchmark},
author={Kenneth Enevoldsen and Isaac Chung and Imene Kerboua and Márton Kardos and Ashwin Mathur and David Stap and Jay Gala and Wissam Siblini and Dominik Krzemiński and Genta Indra Winata and Saba Sturua and Saiteja Utpala and Mathieu Ciancone and Marion Schaeffer and Gabriel Sequeira and Diganta Misra and Shreeya Dhakal and Jonathan Rystrøm and Roman Solomatin and Ömer Çağatan and Akash Kundu and Martin Bernstorff and Shitao Xiao and Akshita Sukhlecha and Bhavish Pahwa and Rafał Poświata and Kranthi Kiran GV and Shawon Ashraf and Daniel Auras and Björn Plüster and Jan Philipp Harries and Loïc Magne and Isabelle Mohr and Mariya Hendriksen and Dawei Zhu and Hippolyte Gisserot-Boukhlef and Tom Aarsen and Jan Kostkan and Konrad Wojtasik and Taemin Lee and Marek Šuppa and Crystina Zhang and Roberta Rocca and Mohammed Hamdy and Andrianos Michail and John Yang and Manuel Faysse and Aleksei Vatolin and Nandan Thakur and Manan Dey and Dipam Vasani and Pranjal Chitale and Simone Tedeschi and Nguyen Tai and Artem Snegirev and Michael Günther and Mengzhou Xia and Weijia Shi and Xing Han Lù and Jordan Clive and Gayatri Krishnakumar and Anna Maksimova and Silvan Wehrli and Maria Tikhonova and Henil Panchal and Aleksandr Abramov and Malte Ostendorff and Zheng Liu and Simon Clematide and Lester James Miranda and Alena Fenogenova and Guangyu Song and Ruqiya Bin Safi and Wen-Ding Li and Alessia Borghini and Federico Cassano and Hongjin Su and Jimmy Lin and Howard Yen and Lasse Hansen and Sara Hooker and Chenghao Xiao and Vaibhav Adlakha and Orion Weller and Siva Reddy and Niklas Muennighoff},
publisher = {arXiv},
journal={arXiv preprint arXiv:2502.13595},
year={2025},
url={https://arxiv.org/abs/2502.13595},
doi = {10.48550/arXiv.2502.13595},
}
@article{muennighoff2022mteb,
author = {Muennighoff, Niklas and Tazi, Nouamane and Magne, Loïc and Reimers, Nils},
title = {MTEB: Massive Text Embedding Benchmark},
publisher = {arXiv},
journal={arXiv preprint arXiv:2210.07316},
year = {2022}
url = {https://arxiv.org/abs/2210.07316},
doi = {10.48550/ARXIV.2210.07316},
}
Dataset Statistics
Dataset Statistics
The following code contains the descriptive statistics from the task. These can also be obtained using:
import mteb
task = mteb.get_task("KinNewsClassification")
desc_stats = task.metadata.descriptive_stats
{
"test": {
"num_samples": 5177,
"number_texts_intersect_with_train": 1294,
"text_statistics": {
"total_text_length": 9500557,
"min_text_length": 137,
"average_text_length": 1835.1471894919837,
"max_text_length": 32815,
"unique_texts": 3367
},
"image_statistics": null,
"audio_statistics": null,
"label_statistics": {
"min_labels_per_text": 1,
"average_label_per_text": 1.0,
"max_labels_per_text": 1,
"unique_labels": 14,
"labels": {
"1": {
"count": 926
},
"10": {
"count": 162
},
"3": {
"count": 408
},
"4": {
"count": 680
},
"0": {
"count": 1311
},
"6": {
"count": 101
},
"2": {
"count": 971
},
"8": {
"count": 108
},
"13": {
"count": 200
},
"9": {
"count": 43
},
"12": {
"count": 78
},
"11": {
"count": 45
},
"5": {
"count": 67
},
"7": {
"count": 77
}
}
},
"hf_subset_descriptive_stats": {
"kinnews_cleaned": {
"num_samples": 4254,
"number_texts_intersect_with_train": 678,
"text_statistics": {
"total_text_length": 7979747,
"min_text_length": 137,
"average_text_length": 1875.822049835449,
"max_text_length": 32815,
"unique_texts": 2686
},
"image_statistics": null,
"audio_statistics": null,
"label_statistics": {
"min_labels_per_text": 1,
"average_label_per_text": 1.0,
"max_labels_per_text": 1,
"unique_labels": 14,
"labels": {
"1": {
"count": 635
},
"10": {
"count": 154
},
"3": {
"count": 366
},
"4": {
"count": 651
},
"0": {
"count": 979
},
"6": {
"count": 87
},
"2": {
"count": 899
},
"8": {
"count": 95
},
"13": {
"count": 133
},
"9": {
"count": 43
},
"12": {
"count": 34
},
"11": {
"count": 36
},
"5": {
"count": 65
},
"7": {
"count": 77
}
}
}
},
"kirnews_cleaned": {
"num_samples": 923,
"number_texts_intersect_with_train": 616,
"text_statistics": {
"total_text_length": 1520810,
"min_text_length": 153,
"average_text_length": 1647.6814734561212,
"max_text_length": 6449,
"unique_texts": 681
},
"image_statistics": null,
"audio_statistics": null,
"label_statistics": {
"min_labels_per_text": 1,
"average_label_per_text": 1.0,
"max_labels_per_text": 1,
"unique_labels": 12,
"labels": {
"1": {
"count": 291
},
"0": {
"count": 332
},
"13": {
"count": 67
},
"12": {
"count": 44
},
"2": {
"count": 72
},
"8": {
"count": 13
},
"4": {
"count": 29
},
"10": {
"count": 8
},
"3": {
"count": 42
},
"11": {
"count": 9
},
"6": {
"count": 14
},
"5": {
"count": 2
}
}
}
}
}
},
"train": {
"num_samples": 20703,
"number_texts_intersect_with_train": null,
"text_statistics": {
"total_text_length": 38259142,
"min_text_length": 33,
"average_text_length": 1847.999903395643,
"max_text_length": 67385,
"unique_texts": 10824
},
"image_statistics": null,
"audio_statistics": null,
"label_statistics": {
"min_labels_per_text": 1,
"average_label_per_text": 1.0,
"max_labels_per_text": 1,
"unique_labels": 14,
"labels": {
"2": {
"count": 3835
},
"4": {
"count": 2685
},
"1": {
"count": 3578
},
"6": {
"count": 408
},
"0": {
"count": 5239
},
"10": {
"count": 668
},
"11": {
"count": 226
},
"3": {
"count": 1770
},
"9": {
"count": 162
},
"8": {
"count": 380
},
"7": {
"count": 301
},
"13": {
"count": 928
},
"12": {
"count": 314
},
"5": {
"count": 209
}
}
},
"hf_subset_descriptive_stats": {
"kinnews_cleaned": {
"num_samples": 17014,
"number_texts_intersect_with_train": null,
"text_statistics": {
"total_text_length": 31840528,
"min_text_length": 33,
"average_text_length": 1871.431056776772,
"max_text_length": 67385,
"unique_texts": 9075
},
"image_statistics": null,
"audio_statistics": null,
"label_statistics": {
"min_labels_per_text": 1,
"average_label_per_text": 1.0,
"max_labels_per_text": 1,
"unique_labels": 14,
"labels": {
"2": {
"count": 3519
},
"4": {
"count": 2578
},
"1": {
"count": 2469
},
"6": {
"count": 352
},
"0": {
"count": 3929
},
"10": {
"count": 628
},
"11": {
"count": 177
},
"3": {
"count": 1548
},
"9": {
"count": 162
},
"8": {
"count": 341
},
"7": {
"count": 301
},
"13": {
"count": 667
},
"12": {
"count": 152
},
"5": {
"count": 191
}
}
}
},
"kirnews_cleaned": {
"num_samples": 3689,
"number_texts_intersect_with_train": null,
"text_statistics": {
"total_text_length": 6418614,
"min_text_length": 117,
"average_text_length": 1739.9333152615884,
"max_text_length": 10925,
"unique_texts": 1749
},
"image_statistics": null,
"audio_statistics": null,
"label_statistics": {
"min_labels_per_text": 1,
"average_label_per_text": 1.0,
"max_labels_per_text": 1,
"unique_labels": 12,
"labels": {
"0": {
"count": 1310
},
"1": {
"count": 1109
},
"3": {
"count": 222
},
"4": {
"count": 107
},
"2": {
"count": 316
},
"10": {
"count": 40
},
"6": {
"count": 56
},
"8": {
"count": 39
},
"5": {
"count": 18
},
"13": {
"count": 261
},
"12": {
"count": 162
},
"11": {
"count": 49
}
}
}
}
}
}
}
This dataset card was automatically generated using MTEB