README.md

---
dataset_info:
  features:
  - name: id
    dtype: int64
  - name: complex
    dtype: string
  - name: simple_reversed
    dtype: string
  - name: simple_tokenized
    sequence: string
  - name: simple_original
    dtype: string
  - name: entailment_prob
    dtype: float64
  splits:
  - name: train
    num_bytes: 115032683
    num_examples: 139241
  - name: validation
    num_bytes: 14334442
    num_examples: 17424
  - name: test
    num_bytes: 14285722
    num_examples: 17412
  download_size: 91848881
  dataset_size: 143652847
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
license: cc-by-sa-4.0
task_categories:
- text2text-generation
language:
- en
pretty_name: MinWikiSplit++
---


# MinWikiSplit++

This dataset is the HuggingFace version of MinWikiSplit++.  
MinWikiSplit++ enhances the original [MinWikiSplit](https://aclanthology.org/W19-8615/) by applying two techniques: filtering through NLI classification and sentence-order reversing, which help to remove noise and reduce hallucinations compared to the original MinWikiSplit.  
The preprocessed MinWikiSplit dataset that formed the basis for this can be found [here](https://huggingface.co/datasets/cl-nagoya/min-wikisplit).

## Dataset Description

- **Repository:** https://github.com/nttcslab-nlp/wikisplit-pp
- **Paper:** https://arxiv.org/abs/2404.09002
- **Point of Contact:** [Hayato Tsukagoshi](mailto:[email protected])

## Usage

```python
import datasets as ds

dataset: ds.DatasetDict = ds.load_dataset("cl-nagoya/min-wikisplit-pp")

print(dataset)

# DatasetDict({
#     train: Dataset({
#         features: ['id', 'complex', 'simple_reversed', 'simple_tokenized', 'simple_original', 'entailment_prob'],
#         num_rows: 139241
#     })
#     validation: Dataset({
#         features: ['id', 'complex', 'simple_reversed', 'simple_tokenized', 'simple_original', 'entailment_prob'],
#         num_rows: 17424
#     })
#     test: Dataset({
#         features: ['id', 'complex', 'simple_reversed', 'simple_tokenized', 'simple_original', 'entailment_prob'],
#         num_rows: 17412
#     })
# })

```

### Data Fields

- id: The ID of the data (note that it is not compatible with the existing MinWikiSplit)
- complex: A complex sentence
- simple_reversed: Simple sentences with their order reversed
- simple_tokenized: A list of simple sentences split by [PySBD](https://github.com/nipunsadvilkar/pySBD), not reversed in order
- simple_original: Simple sentences in their original order
- entailment_prob: The average probability that each simple sentence is classified as an entailment according to the complex sentence. [DeBERTa-xxl](https://huggingface.co/microsoft/deberta-v2-xxlarge-mnli) is used for the NLI classification.

## Paper

Tsukagoshi et al., [WikiSplit++: Easy Data Refinement for Split and Rephrase](https://arxiv.org/abs/2404.09002), LREC-COLING 2024.

## License

MinWikiSplit is build upon the [WikiSplit](https://github.com/google-research-datasets/wiki-split) dataset, which is distributed under the CC-BY-SA 4.0 license.  
Therefore, this dataset follows suit and is distributed under the CC-BY-SA 4.0 license.