From MediaWiki.org
Jump to navigation Jump to search

Other languages:
English • ‎dansk • ‎español • ‎italiano • ‎中文 • ‎日本語
MediaWiki extensions manual
Crystal Clear action run.svg
Release status: beta
MediaInfo Entity Page.png
Description Wikibase extension to manage structured metadata of media files
Author(s) The Wikidata team
Latest version continuous updates
MediaWiki master
PHP 5.5.9+
Database changes Yes
License GNU General Public License 2.0 or later
Translate the WikibaseMediaInfo extension
Check usage and version matrix.
Issues Open tasks · Report a bug

WikibaseMediaInfo is an extension to Wikibase adding a MediaInfo entity type for handling structured data for multimedia files.

The extension hooks into a file description page. It stores supplemental meta data about the file in a MediaInfo entity. The entity's data is displayed on, and is editable from, the File page.


  • In order to use this extension, Wikibase must be installed first. See the basic installation instructions if Wikibase has not been installed yet on your wiki.
  • Download and place the file(s) in a directory called WikibaseMediaInfo in your extensions/ folder.
  • Run Composer to install PHP dependencies, by issuing composer install in the extension directory. (See T173141 for potential complications.)
  • Add the following code at the bottom of your LocalSettings.php:
    wfLoadExtension( 'WikibaseMediaInfo' );
  • Run the update script which will automatically create the necessary database tables that this extension needs.
  • Yes Done – Navigate to Special:Version on your wiki to verify that the extension is successfully installed.

Other extensions[edit]

This extension requires CirrusSearch.

MediaInfo data can be added to a file as it is being uploaded via UploadWizard by setting $wgUploadWizardConfig['wikibase']['enabled'] to true in config.

MediaInfo entities[edit]

A MediaInfo entity is a Wikibase entity stored in a slot on a File page and consists of:

  • an ID in the form Mxxx, where xxx is the id of the associated wiki page
  • any number of captions
  • any number of claims

A caption is a short piece of text plus its language, intended to provide a short description of the file (the same as 'labels' in wikibase).

A claim is a single fact about the file, consisting of a property and a value e.g. Licence=CC-By-SA or Depicts=Dog

(Note that if there is no caption or claim data then the entity is not stored in the database - in this case the entity is known as a 'virtual entity')

More about claims[edit]

The idea of a claim comes from wikibase (for a more detailed explanation see the wikibase data model). A claim consists of a property and value, plus possibly a qualifier.


A property in wikibase describes the data type of a claim, and can be thought of as a category of data about a file. It has a unique id in the form Pxxx (where xxx is a number). Examples of file properties are 'depicts' (what an image is a picture of), 'resolution', 'created by', 'license'.


The value in a claim will very often be a wikibase 'item', which is a concept, topic, or object represented by a unique id in the form Qxxx (where xxx is a number). For example on Wikidata the planet Earth is Q2 and the CC0 licence is Q6938433.


A qualifier is a secondary claim that modifies the primary claim. For example an image might have a tree in the foreground and the sea in the background, in which case it could have 2 'depicts' claims associated with it - 'depicts=tree(applies to part=foreground)' and 'depicts=sea(applies to part=background)'.

How claims are stored[edit]

Claims are stored simply as strings, using the property ids and item ids as appropriate. Using Property and Item ids from Wikidata as an example, an image depicting a black cat could have the claim 'depicts=cat(color=black)' which would be stored as P180=Q146[P462=Q23445]

MediaInfo UI[edit]

MediaInfo entities are shown on, and can be edited from, their associated File page. Captions and claims are shown separately, and claims are split into 'depicts' claims and 'other' claims - this is because for images a 'depicts' claim can be thought of as a special kind of topical "tag", similar to tags on WordPress or flickr.


Search by caption[edit]

Users can search for files by their MediaInfo captions just as they would search for anything else. For example if a user uploads a picture of the Eiffel Tower, and enters ‘Tour Eiffel’ (French) and ‘Eiffel Tower’ (English) as multilingual file captions, the picture is findable by another user searching for either ‘Eiffel Tower’ or ‘Tour Eiffel’.

Search by claim[edit]

Note - examples below all use Property and Item ids from Wikidata.

Searching for a single claim[edit]

To search for a claim, use the haswbstatement keyword. For example, files with the statement ‘depicts Mont Blanc’ can be found by searching for haswbstatement:P180=Q583. Searches for statements can also use qualifiers. For example, to search for images with Mont Blanc (Q583) in the background (Q13217555), a user could use:


Where P518 is the property ‘applies to part’, used as a qualifier

Searching across multiple claims at once[edit]

Claims can be combined using a logical OR in a single search keyword using the pipe character |. For example files depicting a cat (Q146) OR a dog (Q144) can be found using


Claims can be combined using a logical AND by using 2 separate search keywords. For example, files depicting a cat AND a dog can be found using:

haswbstatement:P180=Q146 haswbstatement:P180=Q144

Searching for claims with quantity qualifiers[edit]

To search for a claim with a quantity, use the wbstatementquantity keyword. For example, files that depict 2 humans (Q5) can be found using:


The comparison operators >, >=, < and <= can also be used, so a search for files depicting more than 2 humans can be found using:


Searching for a range of values[edit]

Ranges can be searched for using two wbstatementquantity keywords at once. For example, to find files depicting between 2 and 5 humans (Q5) use:

wbstatementquantity:P180=Q5>=2 wbstatementquantity:P180=Q5<=5

Search implementation[edit]

When the File page is saved, the following MediaInfo data is written to the Elasticsearch index (all examples use Wikidata Property and Item ids):

  • captions data in every language is stored in the opening_text field
  • claims are stored in the format propertyID=value as array elements in the statement_keywords field using the wikibase property ID (and item id, if value is an item)
e.g. ‘depicts house cat’ is stored as P180=Q146
  • claims with qualifiers are stored in the statement_keywords field along with their qualifiers in the format propertyID=value[qualifierPropertyID=qualifierValue]. For example, the Mona Lisa painting (Wikidata item Q12418) depicts a sky (Q13217555) in the background (Wikidata property P518). If we arrange this data in a Wikibase claim, it would be: ‘depicts sky, applies to part background’, which would be stored as P180=Q12418[P518=Q13217555]
Note that claims with qualifiers are also stored without the qualifier, to increase their findability. So, for example, if someone entered the above claim-plus-qualifier, the claim P180=Q12418 is also stored, so that someone can find the file by searching for ‘depicts sky’ alone, as well as by searching for ‘depicts sky, applies to part background’.
  • claims data with qualifiers where the qualifier value is a quantity is stored in the statement_quantity field in the format propertyID=value|quantity, eg. ‘depicts human, quantity 1’ is stored as P180=Q5|1.

Note that not all claims are stored. A claim will be indexed in ElasticSearch only if ALL of the following conditions are true:

  • the claim has a real value (i.e. its value is not ‘no value’ or ‘unknown value’) AND
  • we know how to process its value for indexing. More value processors may be added in future, but currently we require the claim’s value to be either a Q item ID, a string (alphanumeric), or a quantity (numeric) AND
  • the claims's Wikidata property ID is NOT in a configurable list of excluded IDs ($wgWBRepoSettings['searchIndexPropertiesExclude'])AND
  • either
    • its property ID is in a configurable list of property IDs that should be indexed ($wgWBRepoSettings['searchIndexProperties']) OR
    • its property type is in a configurable list of property types that should be indexed ($wgWBRepoSettings['searchIndexTypes'])

Note that for a claim's quantities to be stored, the claim must meet all the criteria above AND the property ID for the quantity qualifier must be present in a configurable list of property IDs ($wgWBRepoSettings['searchIndexQualifierPropertiesForQuantity']).

See also[edit]