PhuongViHuynh/MTHealth

Dataset Card for Multilingual Mental Health QA Dataset

Dataset Summary

This dataset contains multilingual question-answer pairs derived from expert-level mental health resources in Vietnamese, French, English and Korean. The data is split into train, validation, and test sets for each language. Each entry includes a question and a corresponding answer grounded in trusted mental health documents.

This dataset aims to support research in multilingual QA, cross-lingual retrieval, and health-oriented natural language understanding.

Supported Tasks and Leaderboards

• Multilingual Open-domain Question Answering
• Cross-lingual Transfer in Mental Health
• Multimodal QA (if later expanded)

Languages

• Vietnamese (vi)
• French (fr)
• Korean (ko)
• English (en)

Dataset Structure

Each Excel file contains the following columns:

Column Name	Description
STT	Unique STT
id	Unique identifier
question	The question in a given language
answer	The corresponding answer
context	Context extract from source

Example entry:

{
  "id": "fr_0003",
  "question": "Comment gérer les crises de panique ?",
  "answer": "Les crises de panique peuvent être gérées avec des techniques de respiration, de pleine conscience et, dans certains cas, par la thérapie ou la médication.",
  "source": "guide_anxiete_fr.pdf"
}


Contact:
If any links are broken, please get in touch with me to fix!

Phuong Vi Huynh
Email: hpv1812@gmail.com
GitHub: