Dataset · Sanatbek
Aspect-based sentiment analysis of reviews in Uzbek
Sanatbek/aspect-based-sentiment-analysis-uzbek
Annotated restaurant reviews in Uzbek for aspect extraction and sentiment determination.
What's inside
It contains 6,175 sentences from restaurant reviews in Uzbek collected from Google Maps. The annotation follows the SemEval-2014 Task 4 scheme and includes two layers: 7,412 specific aspect terms with positions and sentiment, as well as 7,724 labels across five categories (food, service, atmosphere, and others).
What you can do with it
- Training token-classification models to find mentions of specific dishes and services in Uzbek texts
- Fine-tuning text-classification models to evaluate customer review sentiment by category
- Benchmarking multilingual models in aspect-based sentiment analysis for a low-resource language
- Tasks
- Token classification, Text classification
- Languages
- Uzbek
- Size
- 1K–10K rows
- Downloads this month
- 259
- Likes
- 5
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("Sanatbek/aspect-based-sentiment-analysis-uzbek")