README.md

---
language:
- multilingual
license: cc0-1.0
size_categories:
- 100M<n<1B
pretty_name: Wikidata Entities Connected to Wikipedia
dataset_info:
  features:
  - name: id
    dtype: string
  - name: labels
    dtype: string
  - name: descriptions
    dtype: string
  - name: aliases
    dtype: string
  - name: sitelinks
    dtype: string
  - name: claims
    dtype: string
  splits:
  - name: train
    num_bytes: 4816310500520
    num_examples: 30072707
  download_size: 1844938813418
  dataset_size: 4816310500520
configs:
- config_name: default
  data_files:
  - split: train
    path: data/chunk_*
tags:
- wikidata
- wikimedia
---

# Wikidata Entities Connected to Wikipedia

This dataset is a multilingual, JSON-formatted version of the [Wikidata](https://www.wikidata.org/) dump from September 18, 2024. It only includes Wikidata entities that are connected to a Wikipedia page in any language.
A total of 112,467,802 entities are included in the original data dump, of which 30,072,707 are linked to a Wikipedia page (26.73% of all entities have at least one Wikipedia sitelink).

- **Curated by:** Jonathan Fraine & Philippe Saadé, Wikimedia Deutschland
- **Funded by:** Wikimedia Deutschland
- **Language(s) (NLP):** All Wikidata Languages
- **License:** CC0-1.0

## Dataset Structure

Each row in this dataset represents a Wikidata entity or property as JSON dumps, containing the following:

- ID (QID for entities, PID for properties), which uniquely identifies each Wikidata item or property.
- [Labels](https://www.wikidata.org/wiki/Help:Label), [Descriptions](https://www.wikidata.org/wiki/Help:Description), and [Aliases](https://www.wikidata.org/wiki/Help:Aliases) for the entity in all available Wikidata languages.
- [Sitelinks](https://www.wikidata.org/wiki/Help:Sitelinks) (interwiki links) mapping the entity to pages on [Wikipedia](https://en.wikipedia.org/wiki/Wikipedia), [Wikisource](https://en.wikipedia.org/wiki/Wikisource), [Wikivoyage](https://en.wikipedia.org/wiki/Wikivoyage), and other Wikimedia projects.
- [Claims (statements)](https://www.wikidata.org/wiki/Help:Statements), which are structured assertions pairing a property with one or more values. Claims include additional context such as [qualifiers](https://www.wikidata.org/wiki/Help:Qualifiers), [references](https://www.wikidata.org/wiki/Help:Sources), and [ranks](https://www.wikidata.org/wiki/Help:Ranking).

For more information on the Wikidata's data, visit [Wikidata's help portal](https://www.wikidata.org/wiki/Help:Contents)

## Dataset Creation

### Curation Rationale

The original Wikidata dump represents entity relationships by using claims that only contain property and entity IDs, without including labels. This dataset improves accessibility by adding labels to claims, making it easier to generate textual representations of entities directly, allowing users to read through the dataset sequentially without needing multiple lookups.

### Source Data

[Wikidata](https://www.wikidata.org/) is a free and open knowledge base maintained by Wikimedia Foundation contributors worldwide. It provides structured data for Wikimedia projects like Wikipedia, Wikisource, and Wikivoyage. Unlike Wikipedia, which contains human-readable articles, Wikidata organizes information in a machine-readable format, making it an invaluable resource for NLP applications, knowledge graphs, entity linking, and AI models.
Launched in 2012 by Wikimedia Deutschland, Wikidata has grown into the world’s largest collaboratively edited knowledge graph, containing over 112 million structured data objects. It is maintained by a community of 24,000+ monthly contributors and is available in over 300 languages.

#### Limitations

- The dataset only includes entities linked to Wikipedia, meaning it may exclude certain entities that may be valuable in some contexts.
- Data updates are limited to the September 18, 2024 dump, and changes after this date are not reflected.