From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search
This is a talk page for discussions related to the most recent (Wikidata-based) WikiCite proposal. For discussions on the legacy (pre-Wikidata) proposal see Talk:Wikicite (pre-Wikidata) and Talk:Wikicite (2006 proposal).

Who organizes the post 2016 event series?[edit]

I posted a question at Talk:WikiCite_2016#Who_convened_this_event.3F asking who organized the event. I would appreciate a response there. Blue Rasberry (talk) 12:02, 1 November 2016 (UTC)

I replied there and assume that the post-2016 events are to be organized by roughly the same people, though help from others would of course be welcome. -- Daniel Mietchen (talk) 13:01, 1 November 2016 (UTC)

WikiCite submission(s) to Wikimania 2018[edit]

As per d:Wikidata:Wikimania_2018#WikiCite, we now have a draft doc to coordinate WikiCite-related submissions. About 48h left until the deadline. -- Daniel Mietchen (talk) 00:22, 17 March 2018 (UTC)

Let's write the landing page text![edit]

This is the homepage for the WikiCite project. Unfortunately since the establishment of the project there has been no introductory text published. Without introductory text or a landing page projects seem more closed, newcomers have challenges joining, and the general Wikipedia community cannot develop informed opinions about the project in itself or in the context of its relationship with other Wikimedia projects.

In November 2018 there will be the third WikiCite conference. A goal for this conference should be developing basic documentation about the project. I have some suggested text here -

This text is the kind of information I want to see. It has not gone through the wiki community review process. Someone came to me and suggested that in the lead up to the WikiCite conference, too many people could fail to understand the wiki process and assume that all of this information was the product of consensus. Some people have found this text to be controversial. I posted the text mostly alone except from what I copied and pasted from elsewhere and mixed into this. The alternative to showing this text is continuing with the outreach strategy of the past three years - having almost nothing - which is a problem too, but I agree that "failing to inform" could be better that "incorrectly informing". My real wish is for any text to be here soon, and failing that, for text to come out of the WikiCite conference, and failing that, anything to be posted by March 2019 so as to avoid going another year with silence.

Let's see what WikiCite fans can do to draft text! Blue Rasberry (talk) 13:53, 13 September 2018 (UTC)

@Bluerasberry: Just as an observer, I'd prefer to see a first paragraph that describes what WikiCite is, before jumping into what it can be used for. - PKM (talk) 21:47, 16 November 2018 (UTC)

COI use case[edit]

Because reliable sources have to be independent, I'd really like to see automated linking of sponsors and COIs, similar to that done on PubMed, with this metadata surfaced to editors adding citations and to readers reading them. I've made some COI metadata examples on Wikidata. I've seen way too many Wikipedia articles to fix which cite advertisements formatted to look like journal articles (called "sponsored supplements", and usually under the editorial control of the sponsor). This is a serious and invisible problem on the wikis, and it seems to me as though only a source metadatabase can feasibly fix. I understand from DGG that if WikiCite became its own sister project, it could choose fair-use rules allowing it to host the full texts of COI statements (and abstracts), which might make this task easier. Automated mining of COI statements also seems doable, as it has been done (the software license is unspecified open source, so we could ask about reuse here).

(modified from discussion here) HLHJ (talk) 03:00, 17 November 2018 (UTC)

Citation graph ideas[edit]

A list of cited claims

I'm hacking w/ the Citation Graph designer jndockery at CredCon and think this would make a lovely on-ramp for some flavors of citation browsing. A couple related thoughts worth bookmarking here:

  • Some discussion about interfaces @ wikicite would be great.
  • It might be nice to gather short images and descriptions (on a page here?) of different citation-browsing tools. For instance Scholia provides a wide range of services, one of which is citation browsing.
  • More followups at the conference this month ::
    1. where in the current wikicite pipeline do evaluations of claims, and extracted claims, go?
    2. Should individual claims (and clusters of similar claims) have their own entry, just as citations do?
    3. Should cited sources have trackbacks showing the specific claims cited to them?
    4. Where do people go to clarify or perfect a loose cite into something so precise that a clause in the source is cited to a clause in the target?

SJ talk  04:37, 2 December 2018 (UTC)

Completing the citation graph[edit]

People discussing the broadest view @ WikiCite 2018[edit]

  • BNewbold: universal text and article metadata
  • Konrad F: highly granular citations (source & target), and autogeneration of same
  • {many metadata librarians}
WikiCite poster 2011.pdf

Possibilities and challenges[edit]

  • One page for every cited source [and a sparse-mesh approach to families of similar sources]
  • Handle (common) metadata queries (replace Scopus &c)

Clustering and fuzzy discovery:

  • Cluster similar references [w/ broader/narrower/versioned connections]. work-level alignment.
    • List other work by the same {author, publisher} in similar contexts
  • Graceful degradation of detail (all metametadata, most metadata, much actual content)
  • Disambiguation and classification of names, entities, topics
  • List inbound cites to a source
  • Find everything happening by tag / in a field
  • non-experts can search broadly for clusters of authors, topics, authorities -- if you're new to the field
  • allow fuzzy research questions w/o the researcher needing to figure out which databases to start with.

Clarifications and iteration

  • Classification of citation type / tone
  • Transcludable citation metadata via API
  • Iterative sharpening of citation source and target [until both are a short clause, and you can traverse chains of citation]

Provenance through time and people

  • Tracking species citation through time, w/ descriptions, events, literature: see that extended graph. [which pieces are here or not?]
  • assess authority of author, source, publisher

Capture and index everything

  • allow citing choreographic work and movement data and transmission, through performance
  • persistent identifiers for everything cited

Ease of use: for editors and existing db sync

  • simplify cite generation via WD: comprehensions necessary -- or added on cite? -- to make citing faster.
  • interlayer: use wikicite to dialogue w/ librarians, linking entries in different [biblio] databases
    • the WD data model is not terribly well suited to this purpose. the WD vocabulary should be used, but the data model can be awkward.


  • Walled gardens are being built on top of WD. How do we avoid capture and enclosure?
  • In the same way that incompleteness can be misleading, provenance can be too (when incomplete). e.g. most citations aren't relied on strongly by the article. easy to get prov lineage wrong in a misleading way.

SJ talk  19:38, 28 November 2018 (UTC)

en:Open Database License, a copyleft license, like we use for everything else, has worked for OpenStreeMap. This would also remove moral and legal questions around converting copyleft volunteer-assembled Wikipedia articles into CC-0 data. I saw an article (can't find it now) saying that WD could be used to automatically generate articles of a quality that test readers considered nearly equivalent to that of human-authored articles. I hadn't thought of this as primarily a concern for WikiCite; do you have any examples? HLHJ (talk) 05:59, 12 January 2019 (UTC)