virneo/multilingual-wikihow-qa-16k

---
license: cc-by-nc-3.0
task_categories:
- question-answering
language:
- en
- ru
- pt
- it
- es
- fr
- de
- nl
pretty_name: multilingual-wikihow-qa-16k
size_categories:
- 10K<n<100K
tags:
- wikihow
- QnA
dataset_info:
features:
- name: INSTRUCTION
dtype: string
- name: RESPONSE
dtype: string
- name: SOURCE
dtype: string
- name: METADATA
dtype: string
splits:
- name: train
num_bytes: 144407512
num_examples: 16822
download_size: 76391535
dataset_size: 144407512
---

# Dataset Card for multilingual WikiHow with ~16.8K entries. ~(2-2.2)K for each language.

### Warning [1]

The WikiHow team contacted me and made it clear that **they forbid the use of their data for machine learning purposes**. However, I am not calling for anything, and this dataset only shows the concept, and I strongly advise against violating their ToS.
However, consultation with lawyers made it clear that **dataset can be used for such purposes** if the project has **research purposes**.

### Warning [2]

Source code is kinda **very** bad, and I'm lazy to fix it.

### Dataset Summary

Contains Parquet of a list of instructions and WikiHow articles on different languages.

Each row consists of

* INSTRUCTION
* RESPONSE
* SOURCE (*.wikihow.com)
* METADATA (json with url and language).

### Licensing Information

Data is from WikiHow, license for content is located here:
https://www.wikihow.com/wikiHow:Creative-Commons

### Acknowledgements

This helped me a lot!

https://github.com/HelloChatterbox/PyWikiHow; https://pypi.org/project/pywikihow/