README.md

metadata

task_categories:
  - text-classification
language:
  - vi
pretty_name: Vietnamese Sentiment Analysis Dataset (GPT-4o Annotated)
size_categories:
  - 1K<n<10K

Vietnamese Sentiment Analysis Dataset (GPT-4o Annotated)

Dataset Description

This dataset contains Vietnamese comments/sentences annotated with sentiment labels using GPT-4o. It's derived from the UIT-VSMEC dataset and re-labeled for sentiment analysis tasks.

Dataset Summary

• Source: tridm/UIT-VSMEC
• Annotation Model: GPT-4o
• Language: Vietnamese (vi)
• Task: Sentiment Classification
• License: Inherited from source dataset

Dataset Statistics

Split	Number of Samples
Train	5,548
Validation	686
Test	693
Total	6,927

Sentiment Labels

The dataset uses a 3-class sentiment classification:

• -1: Negative sentiment
• 0: Neutral sentiment
• 1: Positive sentiment

Data Format

Each entry contains:

{
  "sentence": "Vietnamese text here",
  "sentiment": -1 | 0 | 1
}