Hi all! I have a new announcement to make, this time regarding
MediaSearch:[1] research with the new engine will now be easier, since
now also English synonyms can be used to find results.[2]
MediaSearch already leverages Wikidata items to find images that
depict entities that match the search criteria. Now, MediaSearch will
make another step forward, by including also those items that match
English labels or aliases on Wikidata for the original search term. We
expect this to be helpful in discovering media for subjects known
under multiple names (for example, scientific and common names), as
well as in other languages where we often don’t even have
descriptions.
Just to make an example: when a user searches for “bat”, MediaSearch
will also show media where searchable text includes the word
“Chiroptera” (their scientific name, and also an alias on
{{Q|28425}})[3] - and this even if the text doesn’t include “bat”.
Another example would be a Somali user searching for “fiidmeer” (the
Somali word for “bat”) that will get media, through {{Q|28425}},[2]
that include either “bat” or “Chiroptera” - again, even if the text
doesn’t include “fiidmeer”.
For the time being, the feature will include only English labels and
aliases, because of the possible clash in meanings between words
across languages, increasing the odds of false positives. For the same
reason, the current implementation will be very conservative in its
searches, when using aliases.
I am here in case you have any questions or requests for more information.
[1]
https://commons.wikimedia.org/wiki/Special:MediaSearch
[2] Phabricator ticket for reference:
https://phabricator.wikimedia.org/T290652
[3]
https://www.wikidata.org/wiki/Q28425
--
Luca Martinelli [Sannita] (he/him)
Community Relation Specialist