phonsobon/Text_Emotional_kh

Khmer Emotional Text Dataset 🇰🇭

Overview

Khmer Emotional Text Dataset is a public dataset for emotion classification in the Khmer language.

The dataset contains short and medium-length Khmer texts labeled with one emotional category. It is designed for training and evaluating Natural Language Processing (NLP) models such as:

• Sentiment Analysis
• Emotion Classification
• Text Classification
• Transformer Fine-tuning
• Large Language Models (LLMs)

---

Dataset Structure

Each sample contains three fields:

Column	Type	Description
id	int	Unique sample ID
text	string	Khmer text
label	string	Emotion label

Example:

{
  "id": 1,
  "text": "ខ្ញុំសប្បាយចិត្តណាស់ថ្ងៃនេះបានជួបជុំគ្រួសារ។",
  "label": "happy"
}

---

Labels

The dataset contains the following emotion classes:

• happy
• sad
• angry
• fear
• stress
• surprise
• neutral

---

Usage

from datasets import load_dataset

dataset = load_dataset("phonsobon/Text_Emotional_kh")

print(dataset)

Example:

print(dataset["train"][0])

---

Applications

This dataset can be used for:

• Emotion Classification
• Sentiment Analysis
• NLP Research
• Fine-tuning BERT
• XLM-RoBERTa
• mBERT
• Llama
• Gemma
• Qwen
• Khmer NLP

---

Language

Khmer (km)

---

License

Apache License 2.0

---

Citation

@dataset{phon2026khmeremotion,
  author = {Phon, Sobon},
  title = {Khmer Emotional Text Dataset},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/phonsobon/Text_Emotional_kh}
}

---

Author

Sobon Phon

GitHub: https://github.com/phonsobon

Hugging Face: https://huggingface.co/phonsobon