Add wikipedia-2023-redirects dataset

[Abdul-Omira]

Title: Add wikipedia-2023-redirects dataset (redirect resolution + pageviews)

Summary

• New dataset loader: wikipedia_2023_redirects
• Canonical Wikipedia pages enriched with:
  • redirects (aliases pointing to the page)
  • 2023 pageviews (aggregated)
• Streaming support; robust parsing; license notes included
• Tests with tiny dummy data (XML + TSVs); covers streaming

Motivation
RAG/retrieval often benefits from:

• Query expansion via redirect aliases
• Popularity prior via pageviews
  This loader offers a practical, maintenance-light way to access canonical pages alongside their redirect aliases and 2023 pageview totals.

Features

• id: string
• title: string
• url: string
• text: string
• redirects: list[string]
• pageviews_2023: int32
• timestamp: string

Licensing

• Wikipedia text: CC BY-SA 3.0 (attribution and share-alike apply)
• Pageviews: public domain
  The PR docs mention both, and the module docstring cites sources.

Notes

• The URLs in _get_urls_for_config are wired to dummy files for tests. In production, these would point to Wikimedia dumps:
  • XML page dumps: https://dumps.wikimedia.org/
  • Pageviews: https://dumps.wikimedia.org/other/pageviews/
• The schema is intentionally simple and stable. Pageview aggregation is per-title sum across 2023.

Testing

• make style && make quality
• pytest -q tests/test_dataset_wikipedia_2023_redirects.py

Example

from datasets import load_dataset
ds = load_dataset("wikipedia_2023_redirects", split="train")
print(ds[0]["title"], ds[0]["redirects"][:5], ds[0]["pageviews_2023"])

Acknowledgements

• Wikipedia/Wikimedia Foundation for the source data
• Hugging Face Datasets for the dataset infrastructure