license: mit
task_categories:
- text-classification
language:
- twi
tags:
- emotion
- african-languages
- nlp
- text-classification
size_categories:
- 100K<n<1M
This dataset is made available because of Ghana NLP's volunteer driven research work. Please consider contributing to any of our projects on Github
Twi Emotion Analysis Corpus
Dataset Description
This dataset contains emotion-labeled text data in Twi for emotion classification (joy, sadness, anger, fear, surprise, disgust, neutral). Emotions were extracted and processed from the English meanings of the sentences using the model j-hartmann/emotion-english-distilroberta-base. The dataset is part of a larger collection of African language emotion analysis resources.
Dataset Statistics
- Total samples: 432,649
- Joy: 39440 (9.1%)
- Sadness: 29291 (6.8%)
- Anger: 31187 (7.2%)
- Fear: 20131 (4.7%)
- Surprise: 31777 (7.3%)
- Disgust: 34705 (8.0%)
- Neutral: 246118 (56.9%)
Dataset Structure
Data Fields
- Text Column: Contains the original text in Twi
- emotion: Emotion label (joy, sadness, anger, fear, surprise, disgust, neutral)
Data Splits
This dataset contains a single split with all the processed data.
Data Processing
The emotion labels were generated using:
- Model:
j-hartmann/emotion-english-distilroberta-base - Processing: Batch processing with optimization for efficiency
- Deduplication: Duplicate entries were removed based on text content
Usage
from datasets import load_dataset
# Load the dataset
dataset = load_dataset("michsethowusu/twi-emotions-corpus")
# Access the data
print(dataset['train'][0])
Citation
If you use this dataset in your research, please cite:
@dataset{twi_emotions_corpus,
title={Twi Emotions Corpus},
author={Mich-Seth Owusu},
year={2025},
url={https://huggingface.co/datasets/michsethowusu/twi-emotions-corpus}
}
License
This dataset is released under the MIT License.
Contact
For questions or issues regarding this dataset, please open an issue on the dataset repository.
Dataset Creation
Date: 2025-07-04 Processing Pipeline: Automated emotion analysis using HuggingFace Transformers Quality Control: Deduplication and batch processing optimizations applied