hltcoe/megawika-citation-in-progress

MegaWika Citation Source Text

Dataset Access

You can access the dataset using the Hugging Face datasets library:

from datasets import load_dataset
dataset = load_dataset("hltcoe/megawika-citation-text-only")

Dataset Information

MegaWika Citation Source Text is a large-scale, multilingual collection of Wikipedia citations and their source text.

Key Features

• Covers 51 languages
• Contains full text of Wikipedia citations
• Mostly text-only version (little HTML or other markup)

Data Format

Each entry in the dataset is a JSON object with the following structure:

{
  "title": "Title of the Wikipedia article the citation came from",
  "last_revision": "The last revision of the Wikipedia page (not the citation)",
  "url": "URL of the citation",
  "source_text": "The citation source text"
}

Language Files

The dataset is split into multiple files, one for each language. For languages with a large amount of data, the files are further split into parts. Here's a sample of the available files:

• af.jsonl
• ar.jsonl
• az.jsonl
• bn.jsonl
• cs.jsonl
• de.jsonl
• en.split.1.jsonl, en.split.2.jsonl, ..., en.split.8.jsonl
• es.jsonl
• fr.split.1.jsonl, fr.split.2.jsonl, fr.split.3.jsonl, fr.split.4.jsonl
• ...

For a complete list of available files, please refer to the dataset's file structure on Hugging Face.

Usage Examples

Here is an example of how you might use this dataset:

from datasets import load_dataset

dataset = load_dataset("hltcoe/megawika-citation-text-only", "en")["train"]
texts = dataset["source_text"]