Sign in to view source links and access this dataset
Description
A lightweight dataset of (anchor, positive, negative) triplets for training Vietnamese medical text embeddings. Anchors are short section headers, positives are answer snippets from the same article, and negatives are semantically related snippets from other articles in the same category. The dataset was created by mtue29 and last updated on September 25, 2025.
Use Cases
Train text embedding models using TripletLoss based on the provided anchor-positive-negative structure.
Fine-tune models for Vietnamese medical text similarity based on semantically related snippets.
Develop patient education or medical information retrieval systems based on the healthcare domain content.
Strengths
Dataset is specifically designed for contrastive learning with a defined triplet structure.
Negatives are described as 'semi-hard negatives' from semantically related articles, which can improve model training.
The dataset focuses on the Vietnamese language and the healthcare/patient education domain.
Limitations
Description metadata is limited; actual data quality requires manual inspection after download.
Row count is unknown, which may limit suitability assessment.
Column-level documentation is absent; field semantics must be inferred after download.
Provenance
Source
huggingface user mtue29
Freshness
Last updated 2025-09-25 04:35:40; freshness should be verified.
License is unknown; terms of use must be verified before application.