Community Wishlist Survey 2019/Archive/Move references to Wikidata

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search
Random proposal ►

 ◄ Back to Archive  The survey has concluded. Here are the results!


NoN Proposes a social change rather than a technical feature

  • Problem: References in citations are a mere hack. They are not a free text and they do not belong to the article's body. In fact, they are fairly structrured, language independent (the citation is a citation in any language: you just need to control/change the format - which should be done automatically anyway) and consist a piece of information on their own (hence: the standalone bibliographies, bibliographical indices, research desks at libraries etc. exist).
    What is most important, the references as we know them cannot be easily copied between projects / language versions because of different formats of reference citing (stemming from the hack nature). I don't find a reason for keeping redundant references in articles and projects and having them in the main text of the articles (or even worse, in wikicode of templates, infoboxes etc.).
  • Proposed solution: We should try to merge the set of scattered references, have a single database of them and provide all the projects' writers and readers with a single, unified repository of references. Single repository should reduce the redundant work and give a single stop for users (just like Wikimedia Commons is for images, but this one should be much easier to create). This repository would allow the editors to just references to these objects on their local projects, like Polish Wikipedia or French Wiktionary.
As it happens, Wikimedia have already a database/datawarehouse/structured data project: Wikidata. References should be uploaded there as particular items, so they can be reused, easily searched and copied with cut&paste. This needs to be done ASAP, as the current use of references is purely nonsensical. At the moment, the hackish nature of citations generates a dramatic waste of resources in e.g. manhours, makes the re-use difficult and creates an artificial barreer for new editors (who IMHO expect a reference as an object in 2018).
  • Who would benefit: anyone editing Wikipedia and sister projects, Wikidata, people looking just for references, in future: automatic content generators
  • More comments:
  • Phabricator tickets:
  • Proposer: aegis maelstrom δ 09:49, 10 November 2018 (UTC)

Discussion[edit]

  • @Aegis Maelstrom: I included some similar things in my proposal, although I felt it was far too soon to propose actually moving all references to Wikidata. I want this to happen, but there are all sorts of issues (many social) with getting Wikipedia editors to create Wikidata items in order to write articles, including issues with connecting the right author, publisher or publication to the item (especially if the item doesn't exist, or if there are multiple items with the same label – although this could be partly solved by creating new properties for "publisher string" and "publication string"), and with convincing the 100,000 regular Wikipedia editors that they should all partake in doing this – you would have to figure out how to edit Wikidata just to fix a source, and some people just don't need or want that in their lives. The software, data and infrastructure would have to be amazing for it to work at all (inline item search, creation and editing in all three of the existing wikitext editors and in VisualEditor, all publisher items correctly classified and separated from their publications, automatic fixes to items that are blank except for the URL). I'm not sure how realistically achievable it would be, given the vast number of sources used, the smaller but still quite large number of references which aren't formatted correctly, and the even smaller but fairly large number of references which shouldn't even be being used as sources. And you would have to be able to handle page numbers (within articles, presumably), and articles which use alternate citation styles (an extreme example). Jc86035 (talk) 18:17, 11 November 2018 (UTC)
    Hi @Jc86035:, I am aware it will not be supereasy, especially because of the social factor but let's face it: the current way of doing citations is ugly and we cannot be hostages of some small minority who will be willing to do everything in a wikicode. Firstly OFC I would make it a possibility, and yes, the proper solution on WikiData needs to be prepared. This is why I propose to do it in a systematic way (as all the other partial solutions, like translators of citation templates between e.g. language versions of wikis are yet another layer of hacks at best). Badly formatted references are a problem on their own but they are a late phase of migration; in the end I presume there still would be _some_ indices/subtexts in the articles, as not all of them are citations. I have learned about this Wikidata project but it is still quite initial. Anyways, I think this issue is important, needs to be highlighted and solved with proper resources. aegis maelstrom δ 19:19, 11 November 2018 (UTC)
    @Aegis Maelstrom: (I've linked this page from my proposal in a footnote.) After thinking about it a little, I think this (or at least the moving-the-data part) could probably be done within a few months, as long as most Wikipedia editors are on board with it. The things that still seem concerning to me are the pace of software development (which would be a prerequisite to avoid disrupting everyone's workflows massively) and whether Wikipedia editors would actually want it. Jc86035 (talk) 18:16, 12 November 2018 (UTC)

Archived[edit]

Unfortunately, our team will not be able to work on this proposal as it's mostly about wikis' content policies. Thanks for participating in our survey. MaxSem (WMF) (talk) 20:21, 13 November 2018 (UTC)