NEU-ESC: NEU Dataset for Educational Sentiment Analysis and Topic Classification

Overview

NEU-ESC is a dataset designed for Educational Sentiment Analysis and Topic Classification. Each sentence in the dataset is labeled with two attributes:

• Sentiment Analysis: Indicates the sentiment of the text.
• Topic Classification: Categorizes the content into predefined educational topics.

The dataset is collected from online forums, educational social network fan pages, and groups. It has been preprocessed and cleaned to ensure high-quality textual data.

Dataset Statistics

Sentiment Analysis Labels

Label	Sentiment	Count	Percentage	Mean Length
0	Neutral	22,773	69.08%	23.21
1	Positive	4,148	12.58%	24.29
2	Negative	5,250	15.77%	34.30
3	Toxic	845	2.56%	22.78

Topic Classification Labels

Label	Classification	Count	Percentage	Mean Length
0	Spam	405	1.23%	22.71
1	News	902	2.74%	59.55
2	Academic	10,512	31.89%	29.62
3	Other	14,402	43.69%	11.40
4	Service	2,358	7.15%	30.94
5	Jobs & Recruitment	808	2.45%	55.14
6	Personal Affairs	1,478	4.48%	33.17
7	Social Affairs	769	2.33%	67.11
8	Help & Share	670	2.03%	37.03
9	Club & Events	662	2.01%	68.82

Dataset Format

Each sample in the dataset contains:

• Text: The input sentence.
• Sentiment: One of the four sentiment classes (Neutral, Positive, Negative, Toxic).
• Classification: One of the ten topic categories.

Usage

The dataset is useful for:

• Training and evaluating sentiment analysis models in the educational domain.
• Building topic classification models for educational discussions.
• Understanding user engagement in online educational communities.

License

This dataset is released under an open-source license for research and educational purposes. Please ensure proper citation when using it in your work.

Citation

If you use NEU-ESC in your research, please cite:

@misc{neu_esc,
  author = {Nguyen Quang Hung, Mai Phan Quoc Hung, Nguyen Thi Hong Hanh, Duong Phuong Giang},
  title = {NEU-ESC: NEU Dataset for Educational Sentiment Analysis and Topic Classification},
  year = {2024},
  howpublished = {Hugging Face},
  url = {https://huggingface.co/datasets/hung20gg/NEU-ESC}
}

Contact

For any inquiries or issues regarding the dataset, please reach out via Hugging Face discussions or GitHub Issues.