WikiCite: metadata, inbound links, and contextual relevance for every cited work

Underlay
,
1

WikiCite is a spec for a completion of citations:

  • Structured data, inbound links, and contextual relevance for every cited work
  • Entity resolution for {article-citation strings}, {article titles}, {authors}, {affiliations}, {funding sources}

Applied to a growing circle of cites:

  1. scholarly cites on en:wp (almost done),
  2. all scholarly cites + book cites on en:wp,
  3. all scholarly+book cites + all other cites from WP + other major public-knowledge projects,
  4. all cites from the top tier of existing works in wikicite

Current bottlenecks include:

  • Performance (Wikidata as a central repo slows down if you run lots of complex citation queries; all cites are being stored there atm),
  • Access (90% of this metadata exists in digital form, with some duplication, in a dozen private databases and scholarly graphs.
  • Disambiguation (entity resolution + deduplication are unsolved problems)

Groups working on this include:

  • Lens.org (articles and patents)
  • FatCat (articles @ IA)
  • Meta (not yet public)
  • MS Academic, Aminer, GScholar (semi-private data w/ semi-public interfaces, updated regularly)
2

I know @sj already knows about GitHub - centerofci/anot8: Prototype annotation tool for scientific research (PDFs) and https://anot8.org/ . At some point I would like to return to that project and:

  1. Enable anot8.org to save PDF annotations directly to a Solid store / GitHub repo.
  2. Conduct research on how subcitations (which take you to exactly the spot in a research paper) can increase the productivity of researchers who consume them. And that good research etiquette should include uses of tools like hypothes.is and anot8 e.g. this and this.
  3. That to prevent bitrot / linkrot, the web HTMLs and PDFs which are the target of annotations should link to / be backed onto a “permanent” service like archive.org, scihub, etc.
  4. There should be a funding model woven into this to enable longterm support.

Mentioning these potential ideas incase it resonates with any of you fine KnowledgeFutures folk, either now or later :slight_smile: