stract icon indicating copy to clipboard operation
stract copied to clipboard

Structure entities around Wikidata items

Open maxlath opened this issue 9 months ago • 0 comments

Hi! I'm surprised to not find a single mention of Wikidata in the repo as that seems to be the perfect match as a basis for your Entity knowledge graph:

  • It would prepare for internationalization as:
    • you would get links to all Wikipedia editions, not just English
    • Wikidata items come with multilingual labels, descriptions, and aliases that would help you find entities with inputs in many languages
  • It would provide you with entities URLs on most major websites. Example: from Aristotle Wikidata item we can deduce all those URLs, including that StackExchange tag page https://philosophy.stackexchange.com/tags/aristotle

You can easily convert Wikipedia article titles to Wikidata item ids, the latter giving you arguably a stronger guaranty of stability.

You can find Wikidata CC0 dumps here https://www.wikidata.org/wiki/Wikidata:Database_download

Happy to help further if there is interest

maxlath avatar Apr 30 '24 14:04 maxlath