This project is supported by the AI Builder program. The main objective is to classify Thai documents into eight different domains, including:
- 🔮 Imaginative
- 🌱 Natural & Pure Science
- 🔬 Applied Science
- 📚 Social Science
- 🔎 History
- 💵 Commerce & Finance
- 🖌️ Arts
- 🙏 Belief & Thought
-
Source: TNC:Thai National Corpus
-
Organization: Department of Linguistics, Faculty of Arts, Chulalongkorn University
-
After data cleaning, the dataset originally consisting of 45,000 articles has been refined to 36,000 articles
| Sources | Proportion (%) |
|---|---|
| Physical Book | 60% |
| Journal | 25% |
| Newspaper | 5-10% |
| Other publications (e.g. advertising brochures) | 5-10% |
| Online content | <5% |
| Data Split | Proportion (%) | Volume |
|---|---|---|
| Training | 70% | 25,200 |
| Validation | 15% | 5,400 |
| Testing | 15% | 5,400 |