https://phabricator.wikimedia.org/T165653 has some related information, and suggests using Flemish museums as a case study.


On Thu, May 18, 2017 at 8:13 AM Alex Stinson <astinson@wikimedia.org> wrote:
" What would be great to have (but probably doesn't exist) is a case-study showing: here's what happens when you provide metadata, versus here's what happens if you don't, with clear differences in uptake or re-use of media. " 

I am not sure how I would start that research question, there are a number of tactics we could take: 
  • Do analysis of use of media files in Batch uploads, by percent/time: as a success indicator, and then evaluate that against some type of measurement of quality of metadata? 
  • Try to look at the success of "Challenge campaigns" based on different kinds of metadata -- and do some qualitative studies of that work --- asking editors, how it easy it was to use, etc. 
  • Do some type of "Google and Wikimedia Commons search comparison" -- where we do google searches around topics adjacent to upload, and see which GLAM projects get represented in them.
In part, I think we are a bit hampered by the extant metrics tools, to getting a good story here. I had a meeting recently with Giovanni Profeta who has been partnering with WMCH and WMIT to investigate how to visualize the data that we can access: http://www.gprofeta.it/visual-tools-for-glams/index.php?title=Visualizations_to_be_included_in_the_website . Hopefully, this will make its way into WMCH's GLAM metrics tool (once its deployed), but I think some of these visualizations might provide us easier methods of analysis here.

I also think what is missing is a reasonable way to give guidance for and analyze the quality of the metadata to begin with in our context: 
  • Should we develop a metric based on the % of fields used in Artwork/Photograph/Information templates?
  • are there fields that we could prioritize as important? (Creator template, institution template, long vs short descriptions?)
  • Can we describe the use of categories in that metric? 
  • Do we generate this from their upload model or from a random sampling of content? 
I, personally, don't have time to do something like this by myself, until we get closer to the need for GLAM user research for structured commons. However, I would be happy to sit in on/help organizing a working group in this (like we are doing on the documentation-front based in followup  to the conversations at the GLAM-Coordinators meeting). 

Alex 

On Thu, May 18, 2017 at 3:50 AM, Arne Wossink <wossink@wikimedia.nl> wrote:
Thanks all for your feedback; keep it coming if you are aware of any other case-studies!

@Alex, Whatamidoing: those are good and valid points. As a Wikimedian I know that the system works this way, and I know the benefits of providing categories and metadata that are ready for Wikidata. However, when I talk to museums, most of that knowledge is lost. What would be great to have (but probably doesn't exist) is a case-study showing: here's what happens when you provide metadata, versus here's what happens if you don't, with clear differences in uptake or re-use of media. We have anecdotal evidence that this is the case (as shown in this discussion), but that's not always enough to convince GLAMs to support us.


Arne Wossink

Projectleider / Project Manager Wikimedia Nederland

(Werkdagen: maandag, dinsdag, donderdag / Office hours: Monday, Tuesday, Thursday)


Postadres:                                                  Bezoekadres:
Postbus 167                                                Mariaplaats 3
3500 AD  Utrecht                                         Utrecht

2017-05-17 20:34 GMT+02:00 Whatamidoing (WMF)/Sherry Snyder <ssnyder@wikimedia.org>:
Let me expand on what Alex says with two examples of what's great about metadata specifically about how ti integrates with Wikidata:

1)  You can get information about art in general.  Have you read https://blog.wikimedia.org/2016/08/23/wikidata-glam/ ?  There is a map in the middle of the blog post that shows where notable works of art are from, and it was generated from metadata that had been imported to Wikidata.  

2)  Metadata puts your art into Wikipedia articles.  A number of the Wikipedias are using Wikidata-aware infobox templates, which means that when Wikidata has an entry about the artwork, then the artwork or details about it can instantly appear at any Wikipedia that pulls metadata from Wikidata.  Look at the infobox at https://fr.wikipedia.org/wiki/Paysage_de_la_vall%C3%A9e_de_l%27Arno  Everything in that box at the moment is pulled from Wikidata, from the image to the name of the art museum that holds it.  The article itself only contains {{Infobox Art}}.




On Wed, May 17, 2017 at 7:09 AM Alex Stinson <astinson@wikimedia.org> wrote:
Hi Arne, et. al. 

I think the greatest benefit right now, in practical terms, is increasing the discoverability through search and through the current category and linking structured between Wikidata and Wikipedia (for example, interwiki language links on the left-hand side of Wikipedia pages are increasingly including Commons Categories from Wikidata). We don't have a huge amount of evidence, that these things support access (and would welcome any examples folks want to share, like Jos -- or if they have a tactic for examining this data). 

In the long term, the greatest benefit will be ease of migration to Structured Data on Commons -- which has lots of discovery and arbitrary query potential. Recently, I wrote a couple recommendations for Martin Poulter, which, if done with GLAM collections now, I am imagining will help a migration to Structured Data on Commons:  
  • Including as many descriptive metadata templates as you can in existing Commons infoboxes (Institution templates, creator templates, technique templates (basically every type of sub-template type listable at https://commons.wikimedia.org/wiki/Template:Artwork ). These are mapped almost 1-to-1 by a few folks on Commons, so should be fairly easy to migrate long term.
  • Ensuring that Commons categories are mapped one-to-one with how Wikidata concepts are being used (depicts, vs topics related to an object), and doing it on Wikidata with Property:P373 property. 
  • If items are described in Wikidata, adding as many fields as possible. 
The Structured data on Commons team is still being assembled, and will be doing research that builds on some initial research from the Wikidata team before doing community consultations around design next-steps: https://commons.wikimedia.org/wiki/File:HeavyCommonsUserQualitativeResearch.pdf  . I am not sure what the final shape of the project's impacts will be/look like, but the more metadata that is consistently displayed now, the easier it will be for the community or institution to take advantage of the benefits of structured Commons later (such as easing attribution and embedding of the mediafile in other sources, surfacing media files in multilingual search, etc). 

Cheers, 

Alex Stinson

On Tue, May 16, 2017 at 9:32 AM, Arne Wossink <wossink@wikimedia.nl> wrote:
Hi Jos,

Yes, those would be the kind of interaction that would be interesting to see happening as a result of providing metadata to images.

Best,


Arne Wossink

Projectleider / Project Manager Wikimedia Nederland

(Werkdagen: maandag, dinsdag, donderdag / Office hours: Monday, Tuesday, Thursday)


Postadres:                                                  Bezoekadres:
Postbus 167                                                Mariaplaats 3
3500 AD  Utrecht                                         Utrecht

2017-05-16 15:31 GMT+02:00 Arne Wossink <wossink@wikimedia.nl>:
Hi Reem,

Metadata, in this case, refers to the data from the information or artwork (or other) template that's used when an image is uploaded to Commons. So it's not the exif-data, but information about, for example, who's the maker of a painting, when did he make it, what techniques did he use. Take, for example, the data from this upload: https://commons.wikimedia.org/wiki/File:NL-HlmNHA_53004672_Kaaiman.tif

Best,


Arne Wossink

Projectleider / Project Manager Wikimedia Nederland

(Werkdagen: maandag, dinsdag, donderdag / Office hours: Monday, Tuesday, Thursday)


Postadres:                                                  Bezoekadres:
Postbus 167                                                Mariaplaats 3
3500 AD  Utrecht                                         Utrecht

2017-05-14 21:31 GMT+02:00 Jos Damen <josephcmdamen@gmail.com>:
After adding Category:People_being_vaccinated to this file https://commons.wikimedia.org/wiki/File:ASC_Leiden_-_Coutinho_Collection_-_G_07_-_Ziguinchor,_Senegal_-_Vaccination_-_1973.tiff by User Hans Muller, it was picked up by User:Kopiersperre, who added: Category:Jet_injectors and added the image to https://de.wikipedia.org/wiki/Impfpistole 


best regards,

Jos Damen

2017-05-13 15:44 GMT+02:00 Arne Wossink <wossink@wikimedia.nl>:
Hi all,

As best practice we usually encourage GLAMs to provide as many metadata as possible for media donations. However, providing these metadata and "wikifying" them (for examply as part of an upload using Pattypan) can be quite a bit of work, either for a Wikimedian or a GLAM volunteer/staff member.

Do we have any case studies outlining immediate benefits of providing more metadata? For example, does providing more metadata lead to better uptake of images in articles on WP?

Best

Arne Wossink

Projectleider / Project Manager Wikimedia Nederland

(Werkdagen: maandag, dinsdag, donderdag / Office hours: Monday, Tuesday, Thursday)


Postadres:                                                  Bezoekadres:
Postbus 167                                                Mariaplaats 3
3500 AD  Utrecht                                         Utrecht

_______________________________________________
GLAM mailing list
GLAM@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/glam



_______________________________________________
GLAM mailing list
GLAM@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/glam




_______________________________________________
GLAM mailing list
GLAM@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/glam




--
Alex Stinson 
GLAM-Wiki Strategist
Wikimedia Foundation
Twitter:@glamwiki/@sadads

Learn more about how the communities behind Wikipedia, Wikidata and other Wikimedia projects partner with cultural heritage organizations: http://glamwiki.org
_______________________________________________
GLAM mailing list
GLAM@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/glam
--
Sherry Snyder (WhatamIdoing)
Community Liaison, Wikimedia Foundation

_______________________________________________
GLAM mailing list
GLAM@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/glam



_______________________________________________
GLAM mailing list
GLAM@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/glam




--
Alex Stinson 
GLAM-Wiki Strategist
Wikimedia Foundation
Twitter:@glamwiki/@sadads

Learn more about how the communities behind Wikipedia, Wikidata and other Wikimedia projects partner with cultural heritage organizations: http://glamwiki.org
_______________________________________________
GLAM mailing list
GLAM@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/glam
--
Sherry Snyder (WhatamIdoing)
Community Liaison, Wikimedia Foundation