I have no idea how we can find media without having statements on them of the kind "depicts a (some-item)" or "is an instance of (photgraph)", "taken at [Date]", etc., where (items) are represented by Q-something and [values] as usual.
Of course, from "depicts Q112015(=town musicians of Bremen)", we might infer each of "depicts a (donkey)" and "depicts a (dog)" and "depicts a (cat)" and "depicts a (cock)" and likely much more.
Having the bulk of statements on the items depicted, recorded, etc. is imho okay. Yet there may be precision applying only to specific media, such as "a _male_ voice recording of (some-literary-work)". On the long run, I believe, we should have these, too, so as to allow precise queries.
Btw., I agree that the actual location of media files should be of little concern. It is represented by an URL, that is it.
Purodha
"Gerard Meijssen" gerard.meijssen@gmail.com wrote:
Hoi,The use case I was thinking of was to include the images that exist for instance on English Wikipedia. Flickr and other repositories outside the WMF are very much out of scope as far as I am concerned. In my opinion it is silly to associate information about media files with the media file itself. The objective is to search for an image of a "horse" and every image of a "horse" should be included NEVER MIND where the file is "located". When the result is to be restricted to freely licensed images, all images should be included NEVER MIND where the file is "located". NB I love to understand why I am wrong in this.
Thanks, GerardM On 6 September 2014 10:48, P. Blissenbach publi@web.de wrote:Hi
(1) If we want to include media files not on commons, then we shall have to "include" data from foreign sources such as flickr or other types of repositories. We must do so without stealing or damaging the authority of these others. If we connect items to media linking them, or if we assign tags, labels, attributes, etc. to foreign media, or make statements involving them, we can do so of course collaboratively, but we cannot assume other communities to cooperate. Often they will, occasionally they will, not and the latter should not be a hindrance.
(2) Assuming we are incorporating labels, tags and statements (claims) made in other repositories in additioni to simple and obvious technical information, we shall have to decide about incorporating the thesaurii, tagging systems, ontlogies, or whatever they use, first.
(3) Much less complicated imho is the initial step to make files on commons and on other WMF wikis available for searches via WikiData. The goal has to be, imho, that everything we "know" already about them is to be converted into statements and made available to search queries. Since that involves reading descriptions and turning them into statements about media, we get a finer grained categorizing or tagging system than we have today. Itwill automatically become more multilingual as data grows. I currently believe that conversion from existing data has at least partially to be done semiautomatically, likely with suggestor bots, that e.g. ask questions like "Is this cat: o Black, o Brown, o White, o Tigered, ... o Not a cat at all" or "In this sample, you hear the voice of a: o Female, o Male, o Child, o Cannot tell, o Several voices, o No voice at all, ...". That would allow to add considerable volumes missing data in little time, startig from categories existing in the wikis.
(4) Searching should most of the time be a matter of making statements about what you want to find. Basic logical operations need to be availabe so as to limit unwieldy result sets, plus additional stepwise refinements. Semantic Mediawiki or Wolfram Alpha or Library Catalog Search Engines already have many of those ;-)
Purodha
"Gerard Meijssen" <gerard.meijssen@gmail.com[gerard.meijssen@gmail.com]> writes:
Hoi, I am really interested how you envision searching when all those topics are isolated and attached to each file.. I also am really interested to know when you have all those files isolated on Commons, how you will include media files that are NOT on Commons.. This is a normal use case. Thanks, GerardM On 3 September 2014 15:33, James Heald <j.heald@ucl.ac.uk[j.heald@ucl.ac.uk]> wrote:Not really relevant.
The way that this will be achieved will be a "topics" list attached to each file, each topic being a pointer to a Wikidata item.
Sure, Wikidata may be used as one of the sources to help build the topics list; but the topics list will not be on Wikidata, but attached to each file, probably on the CommonsData wikibase.
-- James.
On 03/09/2014 14:28, P. Blissenbach wrote:I strongly support this view: Wikidata should support and ease finding Commons-images. This is not only about proper categorising and tagging in a true multilingual way, but also about determining and assigning various properties - both automatically and manually.
Think for example like an art director creating an image flyer (be it about Wikimania, a national open source movement, or a company) looking for photograhps "predominantly blue" depicing "8 humans or more" of "various ages" in a "neutral or indeterminate environent" and so on, so as to get the hang of it.
Purodha
"Gerard Meijssen" <gerard.meijssen@gmail.com[gerard.meijssen@gmail.com][gerard.meijssen@gmail.com[gerard.meijssen@gmail.com]]> writes:
Hoi, I am firmly opposed to the idea that the Wikidatification of Commons is about Commons. That is imho a disaster.
It is about mediafiles and they exist in many Wikis.
The categories of Commons are in and off themselves useful to a very limited extend. Associating the images they refer to with existing items in Wikidata is one way in which they may be useful. As it is, because of naming conventions and the use of English only, the categories are pretty lame. They do not help me when I am looking for an image in Commons at all.
Really my point is forget about Commons notability start thinking in terms of "what does it take to help people find images". Yes, those people will be 8 years old and they may speak Mandarin or Japanese. Thanks, GerardM
On 3 September 2014 12:05, James Heald <j.heald@ucl.ac.uk[j.heald@ucl.ac.uk][j.heald@ucl.ac.uk[j.heald@ucl.ac.uk]]> wrote:Gerard,
I agree with you that I would like the kind of tools currently available with WikiData also to be available on CommonsData.
Queries that combine the two in an integrated way ought to be made simple and straightforward.
What I don't understand is your objection to placing items that really only have a Commons notability, not a world notability, into a specific namespace, or (notionally) the separate database CommonsData, so that it is possible to run those queries that only relate to Commons information solely on CommonsData, and those queries that only relate to world information solely on WikiData.
Does that not make more sense, than requiring the full bulk of the combined database to always be addressed in order to run any query?
-- James.
On 01/09/2014 07:07, Gerard Meijssen wrote:Hoi, Wikidata is very much a "working database". Its relevance is exactly because of this. Without the connection to the interwiki links, it would not be the same, it would not have the coverage and it would not have the same sized community.
Considerations about secondary use are secondary. Yes, people may use it for their own purposes and when it fits their needs, well and good. When it does not, that is fine too. As it is, we do have all kind of Wiki "junk" in there. We have disambiguation pages, list articles, templates, categories. The challenge is to find a use for them.
When I add statements based on categories, I "document" many categories [1]. As a result over 900 items for categories will show the result of a query in the Reasonator. The results is what I think a category could contain given the subject of a category. For Wikipedians they are articles not categorised, red links and blue links.
There are several reasons why this is not (yet) a perfect fit. The most obvious one is including articles that are not part of the selection eg a list in a category full of humans. Currently not everything can be expressed in a way that allows Reasonator to pick things up in a query.. dates come to mind. Then there are the categories that have an "arbitrary" set of entries.
I am not going to speculate on what kind of qualifiers Commons will come up with. In essence when you can sort it / select it Wikidata will do a better job for you. The "only" thing we have to do is identify the items that fit the mold. This is something that you can often find the basis for in existing categories. Thanks, GerardM
[1]http://ultimategerardm.blogspot.nl/2014/08/wikidata-my-workflow-enriching-wi...]]
http://tools.wmflabs.org/wikidata-todo/autolist.html?q=CLAIM%5B31%3A4167836%...
_______________________________________________ Wikidata-l mailing list Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org][Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org]][Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org][Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org]]] https://lists.wikimedia.org/mailman/listinfo/wikidata-l_____________________...]] Wikidata-l mailing list Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org][Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org]] https://lists.wikimedia.org/mailman/listinfo/wikidata-l%5Bhttps://lists.wiki...]]
_______________________________________________ Wikidata-l mailing list Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org][Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org]]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l%5Bhttps://lists.wiki...]
_______________________________________________ Wikidata-l mailing list Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org][Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org]] https://lists.wikimedia.org/mailman/listinfo/wikidata-l_____________________...] Wikidata-l mailing list Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org] https://lists.wikimedia.org/mailman/listinfo/wikidata-l%5Bhttps://lists.wiki...]
_______________________________________________ Wikidata-l mailing list Wikidata-l@lists.wikimedia.org[Wikidata-l@lists.wikimedia.org] https://lists.wikimedia.org/mailman/listinfo/wikidata-l_____________________... Wikidata-l mailing list Wikidata-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l%5Bhttps://lists.wiki...]