Hoi, I am writing a series of blog posts about Commons. My aim is to identify the issues that I have with how it functions. There are several and I do not bother to write about the ones that are being tackled by the team around Guillaume (as far as it is clear to me what they are doing).
The latest blog is about the difficulty of finding pictures, I am also of the opinion that we have the opportunity to be more of a resource of stock images that are freely licensed. We should stimulate this. Yes Commons is growing rapidly. Its coverage leaves a lot to be desired. In my opinion we need to concentrate on search and coverage to make Commons truly kick ass. Thanks. GerardM
Hi,
Op 28-10-2010 20:56, Gerard Meijssen schreef:
Hoi, I am writing a series of blog posts about Commons. My aim is to identify the issues that I have with how it functions. There are several and I do not bother to write about the ones that are being tackled by the team around Guillaume (as far as it is clear to me what they are doing).
The latest blog is about the difficulty of finding pictures, I am also of the opinion that we have the opportunity to be more of a resource of stock images that are freely licensed. We should stimulate this. Yes Commons is growing rapidly. Its coverage leaves a lot to be desired. In my opinion we need to concentrate on search and coverage to make Commons truly kick ass.
A lot of fun stuff can still be done with the search engine (lucene), but as far as I know there is no development there. Would be nice if the foundation would work on that.
Maarten
Thanks. GerardM
http://ultimategerardm.blogspot.com/search/label/Commons
Commons-l mailing list Commons-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/commons-l
On 28 Oct 2010, at 20:38, Maarten Dammers wrote:
Op 28-10-2010 20:56, Gerard Meijssen schreef:
Hoi, I am writing a series of blog posts about Commons. My aim is to identify the issues that I have with how it functions. There are several and I do not bother to write about the ones that are being tackled by the team around Guillaume (as far as it is clear to me what they are doing).
The latest blog is about the difficulty of finding pictures, I am also of the opinion that we have the opportunity to be more of a resource of stock images that are freely licensed. We should stimulate this. Yes Commons is growing rapidly. Its coverage leaves a lot to be desired. In my opinion we need to concentrate on search and coverage to make Commons truly kick ass.
A lot of fun stuff can still be done with the search engine (lucene), but as far as I know there is no development there. Would be nice if the foundation would work on that.
Regardless of whether the Foundation's interested or not, chapters might be interested in supporting this sort of work (I'm fairly sure that Wikimedia UK would be). I'd recommend that anyone interested in doing such development gets in touch with their local chapter.
Mike Peel
On Thu, Oct 28, 2010 at 10:06 PM, Michael Peel email@mikepeel.net wrote:
On 28 Oct 2010, at 20:38, Maarten Dammers wrote:
Op 28-10-2010 20:56, Gerard Meijssen schreef:
Hoi, I am writing a series of blog posts about Commons. My aim is to identify the issues that I have with how it functions. There are several and I do not bother to write about the ones that are being tackled by the team around Guillaume (as far as it is clear to me what they are doing).
The latest blog is about the difficulty of finding pictures, I am also of the opinion that we have the opportunity to be more of a resource of stock images that are freely licensed. We should stimulate this. Yes Commons is growing rapidly. Its coverage leaves a lot to be desired. In my opinion we need to concentrate on search and coverage to make Commons truly kick ass.
A lot of fun stuff can still be done with the search engine (lucene), but as far as I know there is no development there. Would be nice if the foundation would work on that.
Regardless of whether the Foundation's interested or not, chapters might be interested in supporting this sort of work (I'm fairly sure that Wikimedia UK would be). I'd recommend that anyone interested in doing such development gets in touch with their local chapter.
Actually, we might want to do this in a bit of a more structured way.
Rather than having random people contact their local chapter for that kind of stuff, we could try and have the right people make a "roadmap" of what needs to be achieved, for what purpose etc. Then the chapters (and the Foundation if they wish) could get together and say, ok, this is what resources we can allocate to this, and we'd have a job description made for a roadmap that makes sense and take the best person for the job.
I am pretty sure indeed, that improving commons in that kind of way is something that is "easy to sell" to the general public and most importantly, a huge step forward for the mission.
/me dreams of a really usable Commons ;)
Cheers,
Delphine
Hi guys,
So we're having the tags discussion again? Pops up every once in a while. Tags is a step backwards compared to categories. With tags there is a relation between an object (a photo) and a word. No way of telling what the relation is, what language the tag is, no relations between tags etc etc. So just a word and no more metadata, but it is very easy for the user. On the other side we have semantic web (yes, I said the S word!). That's like an utopia we'll never reach. Categories are somewhere in between. There's a relation between an object and a category, we don't know what that relation is. There are relations between categories, also we don't know what these relations are.
Would be nice if we could use all the information of the current category system to build a better new system. Imho the most important problem of our current system is intersections. Category:Churches gets too crowded so we intersect it with locations (I even wrote a bot to do that). This "hides" a lot of images. We want to add atomic things, let's call them labels. So I want to add the label "church" and the label "Amsterdam" and have some clever software figure out the intersection. Between these labels you can define relations again (maybe even specify more than a relation?) and you can add translations. For the simple user the label should just work like a tag (click, added Amsterdam & Church), but the more advanced user could add more information like translations, relations with other labels, link to Wikipedia articles etc etc making it a powerful system.
Maarten
Hi, In fact, there are so many improvements to do: - first IP should have their interface localised at any time, may be by cookies or browser setting language, I know it's not easy and so one... (there is a bug on bugzilla somewhere) - allow categories to be localised, I know it's not obvious, I know it's difficult for mediawiki and so one, but currently Commons is not easy for non English speakers. (evergreen in many village pumps on Commons)
- add tag system, why not, it could be useful for search. Tag system could be developed like a label one. The first common point are the keywords. These three points could help to make a better Commons
Florian Farge aka Otourly Sur lesprojets wikimédiens et l'Association française,sur OxyRadio, OSM, et sur MOVIM Socio di Wikimedia Italia
________________________________ De : Maarten Dammers maarten@mdammers.nl À : commons-l@lists.wikimedia.org Envoyé le : Ven 29 octobre 2010, 18h 34min 05s Objet : Re: [Commons-l] Use labels in Commons and increase coverage
Hi guys,
So we're having the tags discussion again? Pops up every once in a while. Tags is a step backwards compared to categories. With tags there is a relation between an object (a photo) and a word. No way of telling what the relation is, what language the tag is, no relations between tags etc etc. So just a word and no more metadata, but it is very easy for the user. On the other side we have semantic web (yes, I said the S word!). That's like an utopia we'll never reach. Categories are somewhere in between. There's a relation between an object and a category, we don't know what that relation is. There are relations between categories, also we don't know what these relations are.
Would be nice if we could use all the information of the current category system to build a better new system. Imho the most important problem of our current system is intersections. Category:Churches gets too crowded so we intersect it with locations (I even wrote a bot to do that). This "hides" a lot of images. We want to add atomic things, let's call them labels. So I want to add the label "church" and the label "Amsterdam" and have some clever software figure out the intersection. Between these labels you can define relations again (maybe even specify more than a relation?) and you can add translations. For the simple user the label should just work like a tag (click, added Amsterdam & Church), but the more advanced user could add more information like translations, relations with other labels, link to Wikipedia articles etc etc making it a powerful system.
Maarten
_______________________________________________ Commons-l mailing list Commons-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/commons-l
Op 29 okt 2010, om 18:34 heeft Maarten Dammers het volgende geschreven:
Imho the most important problem of our current system is intersections. Category:Churches gets too crowded so we intersect it with locations (I even wrote a bot to do that). This "hides" a lot of images. We want to add atomic things, let's call them labels. So I want to add the label "church" and the label "Amsterdam" and have some clever software figure out the intersection.
Actually, if tags dont get a hierarchy of themselfs intersecting is one of the features I was planning to build in the Extension (or build in an existing extension if it exists).
One could search for a tag, and another and another narrowing down your search.
I can't imagine how many times I was looking for something simple and being forced to make a specific choise in order to see a picture.
[[Categorie:German scientists who won a nobel price]] Such a category could potentially exist on Wikipedia or Commons.
Instead such a photo could be categorised in: [[Category:Scienco Foobar meeting 2009]] Tag: Scientist, nobel price winner, german
Now for that reason I don't think replacing categories all together is a good thing. Think of categories as sets, pictures that belong together beyond visual similarity.
I mean a tag like "Wikimania 2009" would too specific imho. That's ideal as a gallery and/or category. But how about a category: "Groupphotos taken during Wikimania 2009" ? Too specific. But categorizing images of the event in "Wikimania 2009" and tagging some photos with 'groupphoto' . That would be nice.
On 29 October 2010 17:55, Krinkle krinklemail@gmail.com wrote:
One could search for a tag, and another and another narrowing down your search. I can't imagine how many times I was looking for something simple and being forced to make a specific choise in order to see a picture.
Indeed. Remember that the audience for this is not people interested in ontological structure - they already have the category tree. It's for the casual user to type in a few words they're thinking of. Think something like Getty Images. "Commons is sorta like Getty Images except it's all free and the search sucks."
- d.
Op 29 okt 2010, om 19:35 heeft David Gerard het volgende geschreven:
On 29 October 2010 17:55, Krinkle krinklemail@gmail.com wrote:
One could search for a tag, and another and another narrowing down your search. I can't imagine how many times I was looking for something simple and being forced to make a specific choise in order to see a picture.
Indeed. Remember that the audience for this is not people interested in ontological structure - they already have the category tree. It's for the casual user to type in a few words they're thinking of. Think something like Getty Images. "Commons is sorta like Getty Images except it's all free and the search sucks."
I'll be starting developing this extension on my server this weekend. Just as a basic functioning proof-of-concept to see if there's any major problems I run into. As soon as it's something to look at I'll post and link and see if I can get it in SVN.
Please note that the toolserver link posted earlier is just a static HTML page. It is not a MediaWiki install.
-- Krinkle
in ontological structure - they already have the category tree. It's for the casual user to type in a few words they're thinking of. Think something like Getty Images. "Commons is sorta like Getty Images except it's all free and the search sucks."
And why not work on metadata? If you have possibility of inserting keywords
this will improve information retrieval. Now we have the technical metadata, we need more metadata on the document, we definitely need to improve the descriptive metadata of the represented object. For example, if I have a scan of a public domain book I need both to describe the scan itself and the book. If we would have the possibility of using the Keywords in the DC as tags, in a user-friendly way, we could have both old categories and tags useful for searching. It always go in the direction of Dublin Core (I bet this a recurring discussion too...), and it solves problem of Wikisources too (I'm sorry I repeat myself, but I a cause worth of it ;-))
Aubrey
Hoi, At the Europeana conference in Amsterdam they informed us about a museum where they had not only the terminology used by professionals, they also had labels that were added by the public. Analysis of the positive searches showed that 85% of the searches where because of labels only 15% was because of the official and correct keywords.
This result was discussed in several museums and many museums refused to use labels because there was no truth in the labels. When you want to go dublin core you are talking at best about the 15%.
We need to work on better usage not build another white elephant. Thanks, GerardM
On 29 October 2010 20:30, Andrea Zanni zanni.andrea84@gmail.com wrote:
in ontological structure - they already have the category tree. It's
for the casual user to type in a few words they're thinking of. Think something like Getty Images. "Commons is sorta like Getty Images except it's all free and the search sucks."
And why not work on metadata? If you have possibility of inserting
keywords this will improve information retrieval. Now we have the technical metadata, we need more metadata on the document, we definitely need to improve the descriptive metadata of the represented object. For example, if I have a scan of a public domain book I need both to describe the scan itself and the book. If we would have the possibility of using the Keywords in the DC as tags, in a user-friendly way, we could have both old categories and tags useful for searching. It always go in the direction of Dublin Core (I bet this a recurring discussion too...), and it solves problem of Wikisources too (I'm sorry I repeat myself, but I a cause worth of it ;-))
Aubrey
Commons-l mailing list Commons-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/commons-l
2010/10/29 Gerard Meijssen gerard.meijssen@gmail.com
Hoi, At the Europeana conference in Amsterdam they informed us about a museum where they had not only the terminology used by professionals, they also had labels that were added by the public. Analysis of the positive searches showed that 85% of the searches where because of labels only 15% was because of the official and correct keywords.
This result was discussed in several museums and many museums refused to use labels because there was no truth in the labels. When you want to go dublin core you are talking at best about the 15%.
We need to work on better usage not build another white elephant.
I'm not talking about eliminate the tags/labels, I'm talking about adding (other) proper metadata to our resources, if possible following standards. Tags are an important part of it, but not all. I mean, we can have both the better usage and more reliability.
I don't think that ignoring completely the history and results of librarianship as a science/discipline is a good idea, nor it is see the tags (of DC, for that matter) as a silver bullet. I'm not an expert, but all the digital libraries and repositories I know focus a lot (maybe too much) on metadata, because they guarantee retrievability (not talking about museums here, mainly libraries and collection of texts). We can crowdsource both more traditional metadata and loose but useful tags: why we have to have just one?
Moreover, in the short term you are proably right, and the labels/tags will serve as the main way for findig resources. But in the middle/long term you will lose information, and digital preservation absolutely *needs* metadata. How will you know about a picture if you just have a bunch of useful tags about its content but not the source, creator, publisher, whatever?
Finally, a proper metadata management system (harvesting, disseminating, ecc.) is just a way to *speak* with GLAMs and other projects, like Europeana. Why not go in that direction?
Aubrey
On Fri, Oct 29, 2010 at 9:35 PM, David Gerard dgerard@gmail.com wrote:
Indeed. Remember that the audience for this is not people interested in ontological structure - they already have the category tree. It's for the casual user to type in a few words they're thinking of. Think something like Getty Images. "Commons is sorta like Getty Images except it's all free and the search sucks."
Well said, Sir! Meaningful search is impossible without tags or keywords. And this won't be good only for the general audience, but also for Commoners and Wikipedians. Searching on Commons is a painful experience,a nd tags are the best solution.
On 10/28/2010 2:56 PM, Gerard Meijssen wrote:
Hoi, I am writing a series of blog posts about Commons. My aim is to identify the issues that I have with how it functions. There are several and I do not bother to write about the ones that are being tackled by the team around Guillaume (as far as it is clear to me what they are doing).
I have to admit that I strongly disagree with the blog post
http://ultimategerardm.blogspot.com/2010/10/stimulating-commons-stock-photo....
I think that photoshopped images like that one about Dyslexia have no place anywhere around wikipedia. An image like that just screams "lie", "false" and "designed to manipulate your emotions"; I see that and I think of a cheezy informerical for a phonics program that's going to cure your kid's dyslexia, or some foundation that takes donations to support the lifestyles of the people who run it. It's fundamentally dishonest.
I'm not saying there's no art in that kind of thing, or that it doesn't have a place, but it's not in Wikipedia. If I saw this photo on
http://en.wikipedia.org/wiki/Dyslexia
I'd remove it. In my mind, images used on Wikipedia need to be veridical, which not all commercial illustration is (or needs to be.)
As for the project of "better organizing images" that doesn't necessarily have to be done inside Commons, where a consensus-based culture might inhibit the ability to get things done. I'm taking a crack at it at
That site is nowhere near where I plan it to be in a year, and in the long term it's going to take images in from other sources, but at the moment it's basically a collection of commons images organized a different way. I've got more navigational axes under development.
Hoi, When Chaim Potok wrote about crucifixions in "My name is Asher Lev" he wrote that they are part of the art of a painter. You will find suffering, love, beauty, devotion, belief all expressed in paintings. The western tradition is one where the proverb "a picture paints a thousand words" is accepted.
When you refuse the use of paintings or photoshopped pictures because of the "veridicality", I understand it means truthiness, of pictues I expect you to be the kind of person who does not appreciate how important pictures for many people to have them understand a concept.
When you read the article about dyslexia, there is no picture and the text is not written to explain, it is imho badly written and it can do with an illustration or two. An encyclopaedia is to provide the basic information and getting this information across is what Wikipedia should aim for.
Your requirement of truthiness does not consider what should be primary; do we get the message across and will an illustration help. Thanks, GerardM
On 29 October 2010 16:14, Paul Houle paul@ontology2.com wrote:
On 10/28/2010 2:56 PM, Gerard Meijssen wrote:
Hoi, I am writing a series of blog posts about Commons. My aim is to identify the issues that I have with how it functions. There are several and I do not bother to write about the ones that are being tackled by the team around Guillaume (as far as it is clear to me what they are doing).
I have to admit that I strongly disagree with the blog post
http://ultimategerardm.blogspot.com/2010/10/stimulating-commons-stock-photo....
I think that photoshopped images like that one about Dyslexia have
no place anywhere around wikipedia. An image like that just screams "lie", "false" and "designed to manipulate your emotions"; I see that and I think of a cheezy informerical for a phonics program that's going to cure your kid's dyslexia, or some foundation that takes donations to support the lifestyles of the people who run it. It's fundamentally dishonest.
I'm not saying there's no art in that kind of thing, or that it
doesn't have a place, but it's not in Wikipedia. If I saw this photo on
http://en.wikipedia.org/wiki/Dyslexia
I'd remove it. In my mind, images used on Wikipedia need to be
veridical, which not all commercial illustration is (or needs to be.)
As for the project of "better organizing images" that doesn't
necessarily have to be done inside Commons, where a consensus-based culture might inhibit the ability to get things done. I'm taking a crack at it at
That site is nowhere near where I plan it to be in a year, and in
the long term it's going to take images in from other sources, but at the moment it's basically a collection of commons images organized a different way. I've got more navigational axes under development.
Commons-l mailing list Commons-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/commons-l