---
dataset_info:
  features:
  - name: correct
    dtype: string
  - name: incorrect1
    dtype: string
  - name: incorrect2
    dtype: string
  - name: incorrect3
    dtype: string
  splits:
  - name: train
    num_bytes: 231841100
    num_examples: 3913362
  download_size: 191282565
  dataset_size: 231841100
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---


# Spellchecker Dataset

This dataset contains correct and incorrect spellings of various words in Turkish. It is intended for use in training and evaluating spellchecking models.

## Dataset Structure

The dataset is structured in the following format:
- `correct`: The correct spelling of the word.
- `incorrect1`: A slightly misspelled version of the word.
- `incorrect2`: A moderately misspelled version of the word.
- `incorrect3`: A severely misspelled version of the word.

### Example

| correct  | incorrect1 | incorrect2 | incorrect3 |
|----------|------------|------------|------------|
| bernard  | bernard    | bernrd     | rnard      |
| yemekten | yemekten   | yemektn    | yemekn     |

## Usage

You can load this dataset using the `datasets` library from Hugging Face:

```python
from datasets import load_dataset

dataset = load_dataset("pnr-svc/spellchecker-dataset")