---
license: apache-2.0
language:
- tr
---
# Turkish ABSA Dataset

This repository contains a dataset created for training and evaluating an Aspect-Based Sentiment Analysis (ABSA) model in the Turkish language. The dataset includes sentences with various entities and sentiments.

## About the Dataset

The dataset consists of sentences taken from customer feedback and includes tags related to the entities and sentiments mentioned in these sentences. Each review indicates which entity it is related to and the sentiment towards that entity.

### Columns

- `text`: Sentence taken from customer feedback.
- `entity`: The service or product mentioned in the sentence.
- `sentiment`: The sentiment expressed towards the entity (positive, negative, neutral).

## Usage

You can use this dataset to train and evaluate ABSA models in the Turkish language. Follow the steps below to work with this dataset on the Hugging Face platform.

### Loading

To load the dataset from the Hugging Face platform:

```python
from datasets import load_dataset

dataset = load_dataset("STNM-NLPhoenix/turkish-absa")