Wikidata:Wikimedia Commons/Development

From Wikidata
Jump to navigation Jump to search
Wikidata:Wikimedia Commons

This is a draft about how Wikidata could support Wikimedia Commons in managing both connections with other projects, as well as the basic metadata of the stored medias.

Please use the discussion page to propose changes to this proposal.

Phase 1 (sitelinks) successfully launched on September 23 2013.

Phase 1 (Interlinks)[edit]

Linking from Wikidata[edit]

Direct interwiki links via Wikidata for Commons were enabled on September 23, 2013.[1] Wikidata:Notability has been amended to include items linking to Commons pages such as categories and galleries. The new functionality is only one-directional, reflecting previous interwiki linking practice. A Commons page displays interwiki links to Wikipedia pages but Wikipedia pages don't display the interproject link to the joined Commons page directly. Reverse linking is possible through some properties (Commons category (P373), topic's main category (P910), category's main topic (P301), Commons gallery (P935)) but they are not really functional as interwiki links.

There are no namespace restrictions on linking items with galleries and categories.[2] However Wikidata currently supports linking each item directly to a single Commons page only. So unless and until this changes, a Wikidata item cannot be directly linked to both the relevant Commons category and Commons gallery. The Commons category also cannot be linked to both the article Wikidata item and the category Wikidata item. Interwiki links on the unlinked page then cannot be served by Wikidata. These problems will be present until Wikidata adapts its structure to item logic or Commons changes its linking practices.

Linking from Commons[edit]

Commons categories correspond to both Wikipedia articles and Wikipedia categories. A Commons category page can thus link to either Wikipedia categories (if they exist) or to Wikipedia articles. Links to Wikipedia articles are more common, but sometimes links to Wikipedia categories are preferred.

While every real item should have only one unique Wikipedia article (within one project), unique Wikipedia category (within one project) and unique Commons category, Commons galleries are generally not intended to be unique. A single item can be illustrated by more than a single gallery, perhaps focusing on different aspects of the topic, or just split for ease of navigation (e.g. London Underground stations A-L & M-Z). In some ways, galleries are more like files than articles. Different gallery pages can have interwiki links to the same Wikipedia articles (as can the parent category of the gallery - e.g. commons:Category:Albert Einstein and commons:Albert Einstein). By the way, there are many more categories than mainspace gallery pages, and there have been many more interwiki links from Commons categories to Wikipedia articles than from gallery pages.

Before Commons pages could be linked directly to Wikidata items, the template On Wikidata was created and experimentally used. The template can be enhanced with new functionalities if needed.

Linking from Wikipedias[edit]

Links from Wikipedia articles and Wikipedia categories to Commons (and other sister projects) are not generally implemented through standard interwiki links yet, but through templates such as {{Commons}}, {{Commons category}} and {{Sister project links}} or through various infobox fields.

Bugzilla, proposals, discussions[edit]

Com:category-w:category, com:gallery-w:article[edit]

One possibility would be to limit the linking to one between Wikipedia categories and Commons categories. In other words:

  • ✓ OK Wikipedia article = Commons gallery
  • ✓ OK Wikipedia category = Commons category
  • On hold Wikipedia article = Commons category

Connection can be treated by auxiliary properties as topic's main category (P910) (from the main space item to the category item) and Commons category (P373) (from the main space item directly to the Commons category).

However, this proposal is based on the misapprehension that Commons galleries are equivalents of Wikipedia articles and doesn't help maintain interwiki links from Commons categories to Wikipedia articles.

Com:category-w:article[edit]

Another proposal has been to primarily link Wikipedia articles with Commons categories on Wikidata, with ancillary items created for Wikipedia categories and Commons galleries. For example (with the pages in each column being directly linked in Wikidata):

Sample item Wikipedia categories
(linked through WD interwikis)
Articles/Commons categories
(linked through WD interwikis)
Commons galleries
(linked through WD interwikis)
Commons -/- commons:Category:Pablo Picasso commons:Pablo Picasso
Wikidata (WD) Q9062435 Q5593 Q14916142
Wikipedias (e.g. en) en:Category:Pablo Picasso en:Pablo Picasso -/-

Properties would be used to record linkages across the columns.

Item structure of Wikidata[edit]

Another proposal has been to organize Wikidata primarily by item (not by the type of the linked pages), i.e. one Wikidata item should collect links to both articles and categories relating to that item. Relevant galleries and list articles might also be linked.

This solution would be the most systematic. Regrettably, it was not considered in time (before Phase I) and today would require Wikidata essentially be rebuilt.

Multiple Commons links per item[edit]

A less radical change to Wikidata would be to just allow multiple links to Commons pages from each item. This would enable galleries and articles having the same interwiki links to have these maintained in Wikidata. However, this change doesn't resolve the urge to link to both articles and categories from Commons category pages.

Implementation[edit]

Infoboxes and information sharing[edit]

  • Information from Creator namespace could be transferred from Commons to Wikidata
  • Information from Category:Book templates could be transferred from Commons to Wikidata
  • Category:Institution templates could be transferred from Commons to Wikidata
    • In the other direction, all three of these will require Access to data from arbitrary items (bugzilla:47930) before details can be drawn from the relevant Q-numbers on Wikidata.
  • Campaign pages?
  • Category:License templates may draw information from Wikidata
  • The "information" templates on the file description pages (information about the file and about the item/manifestation) can be replaced by structured data records maintained as subpages or "attachments" to the page.
  • Wiksource draws information from the file description on Commons, such information should be transformed to meta information about the work/manifestation
  • One canonical photo for every item can be suggested at Wikidata. (Wikidata supports the commonsMedia datatype, meaning properties can link to Wikimedia Commons media − see P18 “image”, P51 “audio” and P10 “video”)
  • Wikidata item description can be used as the default intro of Commons category

Media meta data[edit]

Meta data about media files on commons could be maintained as structured data using the same technology that powers wikidata.org. See commons:Commons:Wikidata for media info for a proposal.

Internationalization[edit]

Multilingual Wikidata item labels and item descriptions can help to internalisation of Commons category names, category descriptions and image descriptions and multilingual searching.

Categorization structure[edit]

Prospectively, the categorization structure (relation between categories) can be supported by Wikidata and shared with other projects (primarily with Wikipedias). It can reflect not only simple hierarchy (subcategory/parent category) but also a kind of the relation (hyponymy, meronymy, location, timing, other attribute, agentive and influence relation, modification or derivation, other affiliation etc.) and allow to reference the relations transparently. Maybe, to replace category system with something based on item properties and file properties will be urged.

Currently planned Wikidata phases handle only with basic items. Issue of cross-item categories and meta categories and their interwikis is not solved yet. Some shift from strict modular principle of categorization to some kind of keyword or tag categorization is discussed sometimes (there is a complication on Commons that one file on Commons can have more relevant subjects).

Statistics[edit]

11 August 2014


(274,521)
human: 88,549 (32.3%)taxon: 35,291 (12.9%)administrative territorial entity: 26,001 (9.5%)architectural structure: 38,898 (14.2%)occurrence: 5,168 (1.9%)chemical compound: 286 (0.1%)film: 549 (0.2%)thoroughfare: 5,870 (2.1%)astronomical object: 466 (0.2%)Wikimedia list article: 956 (0.3%)Wikimedia disambiguation page: 678 (0.2%)artificial object: 16,945 (6.2%)other P31/P279: 20,575 (7.5%)no P31/P279: 34,289 (12.5%)
  •   human: 88,549 (32.3%)
  •   taxon: 35,291 (12.9%)
  •   administrative territorial entity: 26,001 (9.5%)
  •   architectural structure: 38,898 (14.2%)
  •   occurrence: 5,168 (1.9%)
  •   chemical compound: 286 (0.1%)
  •   film: 549 (0.2%)
  •   thoroughfare: 5,870 (2.1%)
  •   astronomical object: 466 (0.2%)
  •   Wikimedia list article: 956 (0.3%)
  •   Wikimedia disambiguation page: 678 (0.2%)
  •   artificial object: 16,945 (6.2%)
  •   other P31/P279: 20,575 (7.5%)
  •   no P31/P279: 34,289 (12.5%)
Module:Statistical data/by project/classes, 2015-07-18

Notes[edit]

  1. mailarchive:wikidata-l/2013-September/002892.html
  2. mailarchive:wikidata-l/2013-September/002834.html
  3. mailarchive:wikidata-l/2013-September/002892.html