Datasets:
metadata
license: apache-2.0
language:
- vi
task_categories:
- text-classification
tags:
- sentiment-analysis,
- absa,
- fashion-reviews,
- fashion
configs:
- config_name: default
data_files:
- split: train
path: FashionReviews.csv
Dataset Card: Vietnamese Fashion Reviews Aspect-Based Sentiment Analysis
Dataset Summary
This dataset comprises customer reviews of fashion products, with all review texts exclusively in Vietnamese. Sourced from a diverse range of e-commerce platforms and repositories—including Shopee, Lazada, Kaggle, and Amazon Fashion Reviews—the collection offers a rich and varied representation of consumer feedback. It is specifically designed and curated for Aspect-Based Sentiment Analysis (ABSA) tasks, enabling the accurate classification of customer attitudes toward distinct product attributes such as material, style, size, price, and service.
Data Visualization
| Before data augmentation | After data augmentation |
|---|---|
 |
 |
Data Structure
Label mapping (Applicable to all aspect columns):
0: Không đề cập (Not mentioned)1: Tiêu cực (Negative)2: Trung lập (Neutral)3: Tích cực (Positive)
The table below is about the structure of the dataset:
| Column Name (Vi) | Data Type | Description |
|---|---|---|
| STT | int64 | Sequential index or ID of the review |
| Nội dung review | string | The text of the customer's review |
| Chất liệu | int64 | Sentiment regarding the fabric, material quality, and physical feel of the clothing |
| Kiểu dáng | int64 | Sentiment regarding the design, aesthetics, color, and fashionability of the product |
| Kích cỡ | int64 | Sentiment regarding the fit, sizing accuracy, and how well it matches body measurements |
| Giá cả | int64 | Sentiment regarding the cost, affordability, and overall value for money |
| Dịch vụ | int64 | Sentiment regarding customer support, shipping speed, packaging, and seller communication |
Contributors
Châu Quốc Vinh
Vũ Trọng Nghĩa