On Wikimedia Commons, a lot of metadata (including license and author) are not machine readable. There is an API module, iiprop=extmetadata which can be used to retrieve some values (example), but as the information is entered as free text into the file description page itself, this is not perfect. The ongoing Structured data on Commons project aims to move the metadata as fully-structured data and will eventually supersede the machine-readable data presented in this page.

In the meantime, and to ease a transition towards more structured data at a future time, Wikimedia Commons use a set of standard templates which have been made machine-readable in some ways, through HTML elements. Some scripts already make use of that. It is worth noting that this data is available for any wiki using Wikimedia Commons, where it can be read from the html of the File: page just as other local data.

Machine readable data

Machine readable data set by infobox templates

These are several standard infobox templates tagging different elements of the template with different tags to allow parsing of the information. Several different styles of tags are used:

Alternative format for CommonsMetadata

Because the table + id based format proved very hard to add to templates which were not formatted similarly to the Commons information template, CommonsMetadata allows an alternative format, similar to license templates: the whole information template has to be enclosed in a fileinfotpl class and the tag containing the specific information needs to have a fileinfotpl_* class (same names as above, but class, not id).

Machine readable data set by license templates

Introduced in October 2010, using classes <span class="licensetpl_XXX">

licensetpl

An element identifying a license. Wraps the entire license code and should be a SINGLE license, not a multi license.

licensetpl_short

Short name of the license: “Public domain”, “CC BY-SA 3.0”, “CC by 2.0 fr”, etc.

licensetpl_long

Long name of the license: “Public domain”, “Creative Commons Attribution-Share Alike 3.0”,

licensetpl_attr_req

Whether attribution is required. “true” or “false”.

licensetpl_attr

The requested attribution: Free text.

licensetpl_link_req

Whether a link to the license is required for this license. “true” or “false”.

licensetpl_link

The link to the license deed. “www.creativecommons.org/licenses/by-sa/XXX/YYY”

licensetpl_nonfree

“true“ if this is a non-free license (not used on Commons, only on wikis with an EDP)

Multiple licensetpl blocks for the same work might be wrapped in a block using the class licensetpl_wrapper.

Templates setting this information

• Templates setting licensetpl include:

Machine readable data set by style formatting templates

Style formatting templates, meant to provide uniform styles to different families of non-license templates, carry machine readable data identifying these families.

Machine readable data set by non-copyright restriction templates

Templates regarding non-copyright legal restrictions carry these classes to identify specific types of restrictions.

Machine readable data set by specific templates

More machine-readable data are set. Here is a non-exhaustive list:

Machine-readable data set by location templates

{{Location}} and similar templates add machine-readable geocodes in the following format: <span class="geo">12.34;24.68</span> (latitude and longitude as floating-point numbers, separated by a semicolon). The coordinates use the en:WGS84 system (same as the GPS and most online maps). See Commons:Geocoding for more details.

Usage

MediaWiki API

The MediaWiki API now serves a limited number of metadata. Consider the following query:

(Open in API Sandbox) that returns some useful parameters such as Credit, Artist, LicenseUrl and Copyrighted and is used by Media Viewer, for example.

Scripts using machine-readable data

External tools

Defining new machine readable data

• Do NOT use HTML id's, use classes. An ID can only be used once per page and most of these fields can occur multiple times per page. Consider for instance descriptions of derivative works, which can include information about the original and the derivative.

• When possible, wrap the actual data, not some field header. This last method is historically used for all our Information templates, but much harder to support in the long run.

• Wrap data, not the way the data is formatted.
• Expect that formatting is lost when converting to data. Visual dress up is not part of the information.

• Don't wrap multiple units of information inside one field. There is a difference between a publication date and a creation date. Both are dates, but both are different 'data fields'. Also CC BY-SA-4.0-3.0-2.5 is not a license name, those would be 3 licenses with the name CC BY-SA-##.
• Make sure that the data value has one unit, or outputs one consistent unit.

Problems

There are a few things that are currently NOT or badly recognizable. These include:

• Derivative works
• Works included in works. See also Category:FoP_templates
• licenses derivates or works included in works are a mess.
• Author vs. Copyright holder
• usernames vs 'real names'
• Catalogue IDs etc
• VRTS permissions
• Publication date vs creation date

• Donating institutions of materials
• Anything that is NOT using the above structures is not recognizable at all and will require manual cleanup at some point.
• Heirs: {{Heirs-license}}
• Multilicensed CC works, that use {{Cc-by-3.0,2.5,2.0,1.0}}, {{Cc-by-sa-2.5,2.0,1.0}}, {{Cc-by-sa-4.0,3.0,2.5,2.0,1.0}} or {{Cc-by-all}}.
• Non-licensed works: {{Copyrighted free use}}, {{Attribution}} (Problem, how to describe this grant of rights success ?)
• Improvised File description templates like User:Tevaprapas/Information
• Templates denoting the copyright of partials of the work: {{Copyright information}}