Page MenuHomePhabricator

Add P6243 (digital representation of) to MediaSearch algorithm
Closed, ResolvedPublic

Description

We have this:
The current MediaSearch algo only uses one structured data statement: P180 (depicts). However, files with artwork can use a separate property, P6243 (digital representation of).

We want this:
Include P6243 in the MediaSearch algorithm

Acceptance Criteria:

  • An indication that this is working properly - Searches for "the kiss" should return and prominently rank these results

During development, please test the following:

  • Test this feature while logged in AND logged out
  • Test this feature on at least one mobile browser

Related Objects

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptOct 7 2020, 6:25 PM
Ramsey-WMF updated the task description. (Show Details)Oct 7 2020, 6:26 PM

Change 633978 had a related patch set uploaded (by Matthias Mullie; owner: Matthias Mullie):
[mediawiki/extensions/WikibaseMediaInfo@master] Allow other properties to be searched

https://gerrit.wikimedia.org/r/633978

I suspect we want to rank 'digital representation of' slightly better than 'depicts' matches (given that it's more specific)?

Change 633982 had a related patch set uploaded (by Matthias Mullie; owner: Matthias Mullie):
[operations/mediawiki-config@master] Add another SDC property to search for matching media statements

https://gerrit.wikimedia.org/r/633982

Change 633978 merged by jenkins-bot:
[mediawiki/extensions/WikibaseMediaInfo@master] Allow other properties to be searched

https://gerrit.wikimedia.org/r/633978

Change 633982 merged by jenkins-bot:
[operations/mediawiki-config@master] Add another SDC property to search for matching media statements

https://gerrit.wikimedia.org/r/633982

Mentioned in SAL (#wikimedia-operations) [2020-10-29T11:12:49Z] <urbanecm@deploy1001> Synchronized wmf-config/InitialiseSettings.php: 28152b7387082b79d71cfbf28be740ffe629ee50: Add another SDC property to search for matching media statements (T264925) (duration: 00m 58s)

I verified this code is on production (via a search for Atlas van Dirk van der Hagen, with a result that could only be found via that text because it has a corresponding P2643 statement). However, we're still not getting the desired results for "The Kiss". Thoughts, @matthiasmullie ? 🤔

Yeah looks like it's working. The second result (File:The Kiss - Gustav Klimt - Google Cultural Institute.jpg) is one of P2643=Q698487 and is largely in that position because of that statement.
A lot of others are being drowned out by many other file with (title/description) occurrences of "kiss", though - that balance is another problem (that we're currently looking into!)

matthiasmullie closed this task as Resolved.Tue, Nov 10, 9:32 AM