Fixing Wikidata (by viewing it as a series of tables)

From semantic-mediawiki.org
SMWCon Fall 2023Fixing Wikidata (by viewing it as a series of tables)
SMWCon Fall 2023
Fixing Wikidata (by viewing it as a series of tables)
Talk details
Description: Thoughts on how to improve Wikidata (and yes, Wikibase) querying and editing.
Speaker(s): Yaron Koren
Slides: see here
Type: Talk
Audience: Everyone
Event start: 2023/12/13 14:45:00
Length: 30 minutes
Video: click here
Keywords:
Give feedback

Wikidata, now 11 years old, is an amazing site, holding probably the largest collection of general-knowledge structured data ever created, with over 1 billion statements. It also powers the infoboxes of many - or even most - of the language Wikipedias.

However, despite all of those accomplishments, there are still areas where it falls short of its potential:

  • It seems to be barely used as a data source among non-Wikipedia sites and applications, wiki or otherwise
  • Tied in with that, no LLMs seem to use it as a source
  • Querying of Wikidata by individuals seems minimal also
  • The data within Wikidata is in places relatively "messy" (which is no discredit to the enormous amount of work that volunteers have put into it)


It is my belief that some or all of these problems can be improved by viewing Wikidata less as a free-form graph and more as a structured set of tables - with a more structured approach to editing and querying. This talk will present some interfaces that take advantage of this approach - both real (such as Wikidata Walkabout) and theoretical.

The talk will also cover the strange rise of Wikibase, and how its interface could be improved in similar ways, for people who really want to use it.