In response about the discussion on wikien-l about fair use and placeholder images for biographies (and other things?) I have written a new tool, WatchFlickr.
Given a category (at selectable depth), it retrieves all articles in that category, checks if it lacks an image, and if so, searches Flickr for relevant CC-BY and CC-BY-SA images. It shows found flickr thumbnails, links to the wikipedia and flickr page, and to Flominator's Flinfo tool.
Not that due toolserver problems, several wikipedias, notably en, are not available to search right now. German wikipedia works, though, in case you want to test.
URL: http://tools.wikimedia.de/~magnus/watchflickr.php
I have already found, uploaded, and used about a dozen "celebrity" images, both adding a new image and (on en.wp) replacing fair use and placeholder images.
Cheers, Magnus
On 10/1/07, Magnus Manske magnusmanske@googlemail.com wrote:
In response about the discussion on wikien-l about fair use and placeholder images for biographies (and other things?) I have written a new tool, WatchFlickr.
WOW! That is genius. I just gave it the French department of Finistere, and it came up with some really good photos. I would love to see tools like this integrated better into Wikipedia, so that everyone can know about and use them.
Given a category (at selectable depth), it retrieves all articles in
that category, checks if it lacks an image, and if so, searches Flickr
An option to include pages that already have one or more images would be good.
for relevant CC-BY and CC-BY-SA images. It shows found flickr
thumbnails, links to the wikipedia and flickr page, and to Flominator's Flinfo tool.
Not that due toolserver problems, several wikipedias, notably en, are not available to search right now. German wikipedia works, though, in case you want to test.
In what way doesn't the en "work"? Is it just running off an old dump? Or when I search "en" is it actually searching "de"?
Steve
Thanks for the parise :-)
On 10/3/07, Steve Bennett stevagewp@gmail.com wrote:
An option to include pages that already have one or more images would be good.
I'll add that.
In what way doesn't the en "work"? Is it just running off an old dump? Or when I search "en" is it actually searching "de"?
The server that replicates en for the toolserver apparently went online again tonight, so en's now working as well.
Cheers, Magnus
Seriously, this tool is phenomenal, and when linked with Brian's upload tool, provides a straightforward way to get new, relevant images into Wikipedia. I really hope you'll spread the word on Wikipedia. Perhaps mention it to the Signpost guy?
http://en.wikipedia.org/wiki/Spotswood%2C_Victoria looks much better with two new images, don't you think?
Some other appearance suggestions: * Rename the links that appear next to the image. I suggest making the image itself a link to the flickr page (and hence unlinking the title), then having wording like: Upload with [[Bryan]] / [[flinfo]]. Actually I don't really understand how flinfo works, so perhaps you could just decide which is the better method and only link to that. * Show the flickr tags for each image. That would help a lot in determining if the thing in question is referring to the actual subject. I'm looking at suburbs, and very often there's a place in the US or UK with the same name...the tags would help here.
I do note though, that it's still a *lot* of steps. Most if it's pretty obvious and just a question of clicking the right link, but I wonder if you can streamline it still further.
Let's see, once you've done your search, it's: 1.Middle click the title of an image to view it in another tab. Decide whether it's appropriate. 2.Return to main tab (if it's not, close the tab) 3. Open the "Brian's Upload Tool" link in a new tab 4. Click the Continue button 5. Type categories, adjust description and filename if necessary. (sometimes this page displays wrongly in my browser) [most of this step seems redundant, as you can change all the information except the filename on the next page anyway] 6. Click Continue 7. Click Save Page 8. Click "Flickr Uploader" link. 9. Click "Comment" link 10. Type a comment, usually referring to the page you'll be adding the image to. Tense can be complicated as the image isn't actually visible *yet*. 11. Close the tab and return to your main tab. 12. Click the Edit link next to the original article, opening in a new tab. 13. Type the name of the image somewhere, either as part of an infobox, or an [[image:...]] link. Usually involves flipping back somewhere to get the exact image name. 14. Type an edit summary (usually "Add image") 15. Click save. 16. Check that everything looks ok. 17. Close the tab.
And that's basically the shortest possible version. Going further would include making a proper category on Commons (which is itself a subcategory of something), possibly making an "article" on Commons, then linking to either the category or the article page from the Wikipedia article via the {{commons}} or {{commonscat}} templates. You could also go through the other language editions, adding the image there too.
Conclusion: It's still a lot of clicking and typing, and feels very "manual". How could this be better? How could *Media*Wiki better integrate images and text?
Let me dream a moment: 1. Somehow, categories on all the wikipedias and commons are linked. 2. By default, articles display up to N images from the appropriate Commons category, without being explicitly told which images or which category. Say N=10. Images are tagged so that MediaWiki can choose which image should be the "main" image, and which should be shown in a Gallery section. 3. With a single click, the flickr image is prepared for upload to Commons with appropriate categories. You simply tweak image name and description, then press save. 4. The Wikipedia page (or pages) is then shown with the new image for approval. Since this is automatic, you simply have to check that there aren't any special issues to know about.
Is this desirable? Is it feasible? What are the downsides?
Steve
On 10/4/07, Steve Bennett stevagewp@gmail.com wrote:
Some other appearance suggestions:
- Rename the links that appear next to the image. I suggest making the image
itself a link to the flickr page (and hence unlinking the title), then having wording like: Upload with [[Bryan]] / [[flinfo]]. Actually I don't really understand how flinfo works, so perhaps you could just decide which is the better method and only link to that.
- Show the flickr tags for each image. That would help a lot in determining
if the thing in question is referring to the actual subject. I'm looking at suburbs, and very often there's a place in the US or UK with the same name...the tags would help here.
Both done. :-)
I do note though, that it's still a *lot* of steps. Most if it's pretty obvious and just a question of clicking the right link, but I wonder if you can streamline it still further.
Let's see, once you've done your search, it's: 1.Middle click the title of an image to view it in another tab. Decide whether it's appropriate.
You can actually use the left button, it will open in a new window/tab, depending on your browser settings. Do you have a better way to preview the image? Maybe some JavaScript expansion thing within the tool page itself?
2.Return to main tab (if it's not, close the tab) 3. Open the "Brian's Upload Tool" link in a new tab 4. Click the Continue button 5. Type categories, adjust description and filename if necessary. (sometimes this page displays wrongly in my browser) [most of this step seems redundant, as you can change all the information except the filename on the next page anyway] 6. Click Continue 7. Click Save Page 8. Click "Flickr Uploader" link. 9. Click "Comment" link 10. Type a comment, usually referring to the page you'll be adding the image to. Tense can be complicated as the image isn't actually visible *yet*. 11. Close the tab and return to your main tab. 12. Click the Edit link next to the original article, opening in a new tab. 13. Type the name of the image somewhere, either as part of an infobox, or an [[image:...]] link. Usually involves flipping back somewhere to get the exact image name. 14. Type an edit summary (usually "Add image") 15. Click save. 16. Check that everything looks ok. 17. Close the tab.
So, just to clarify, steps 4-17 are actually the upload tool :-)
And that's basically the shortest possible version. Going further would include making a proper category on Commons (which is itself a subcategory of something), possibly making an "article" on Commons, then linking to either the category or the article page from the Wikipedia article via the {{commons}} or {{commonscat}} templates. You could also go through the other language editions, adding the image there too.
Conclusion: It's still a lot of clicking and typing, and feels very "manual". How could this be better? How could *Media*Wiki better integrate images and text?
Let me dream a moment:
- Somehow, categories on all the wikipedias and commons are linked.
Hmm. Categories on wikipedia are about articles, categories on commons are about images. Even if we could somehow link the two, most images in the commons category would only apply to few articles in the wikipedia category.
Note that it might be easy to semi-automatically link categories between wikipedias: * Get all articles in a category on xx.wikipedia * For each of these, find the language link to yy.wikipedia, and list the categories of that article * sort categories at yy.wikipedia by number of occurrence * the most occurring category is most likely to be the same as the original one on xx.wikipedia If this is desparately wanted, I might write Yet Another Tool ;-)
- By default, articles display up to N images from the appropriate Commons
category, without being explicitly told which images or which category. Say N=10. Images are tagged so that MediaWiki can choose which image should be the "main" image, and which should be shown in a Gallery section.
So, which category to chose for [[en:Horse-ripping]]? [[Category:Horses]]? That would be linked to commons [[Category:Horses]]. I don't see any images of ripped horses there (and I'm rather glad about that:), just images of horses, and showing 11 images of plain horses is rather pointless in [[en:Horse-ripping]]. Even a single image of a horse does not really make a lot of sense there, IMHO.
- With a single click, the flickr image is prepared for upload to Commons
with appropriate categories. You simply tweak image name and description, then press save.
So the commons categories that match the (say) en categories of the article for which flickr images were found are added to the uploaded flickr image? Sounds good. But isn't that what CommonSense [1] does?
- The Wikipedia page (or pages) is then shown with the new image for
approval. Since this is automatic, you simply have to check that there aren't any special issues to know about.
That would be the end point of any (manual or automated) upload process anyway, right?
Counter-suggestion: Have a commons upload tool that will 1. Take the flickr URL and optionally a new filename 2. Check if the flickr image has a suitable license 3. Run CommonSense (or the like) to find appropriate categories on commons 4. Upload the image to commons under the appropriate bot name, using the new filename, flickr template, and categories 5. Open the final image page and present it to the user (the other steps are "invisible")
Alternatively, if my "upload from URL" extension were enabled on commons, it could fill in the entire upload form. The user could check the wikitext before uploading, and upload under his own name with a single click.
Cheers, Magnus
[1] http://tools.wikimedia.de/~daniel/WikiSense/CommonSense.php
OK, I went ahead and wrote a flickr-to-commons upload tool myself...
WatchFlickr has an extra line per image, with a pre-filled image name suggestion and an upload button. This will launch the upload tool.
The tool will check for valid license again, get metadata from flickr, get categories from CommonSense, and upload it to commons as "User:File Upload Bot (Magnus Manske)".
You will see the progress, and the wiki text it uses as upload description. Currently, you will have to click one more link to go to the new image page.
I am not sure what happens if you try to overwrite an existing image. The tool should probably check and auto-rename the image. Something to do for tonight...
Yes, I realize the potential abuse. However, I see these as pretty limited, compared to "let anyone upload any image". It is probably easier to create a new user account and upload stuff through that than to * create account on flickr * upload stuff to flickr * mark it as CC-BY * run my new upload tool
I think the advantages (easy upload) outweight the risks by far. Worst case, block my bot...
Cheers, Magnus
On 10/5/07, Magnus Manske magnusmanske@googlemail.com wrote:
You will see the progress, and the wiki text it uses as upload description. Currently, you will have to click one more link to go to the new image page.
Much better! However, it would be better still if it automatically took you to editing the new image page.
Steve
On 10/4/07, Magnus Manske magnusmanske@googlemail.com wrote:
Yes, I realize the potential abuse. However, I see these as pretty limited, compared to "let anyone upload any image". It is probably easier to create a new user account and upload stuff through that than...
I don't see how it's any more dangerous than Erik's FlickrLickr bot. Also we have a handful of prolific IP editors who, for whatever reason, refuse to create an account. This may be of interest to them.
—C.W.
On 10/4/07, Magnus Manske magnusmanske@googlemail.com wrote:
Do you have a better way to preview the image? Maybe some JavaScript expansion thing within the tool page itself?
Possibly - even a frame or something. Bigger thumbnails would also help.
2.Return to main tab (if it's not, close the tab)
- Open the "Brian's Upload Tool" link in a new tab
- Click the Continue button
- Type categories, adjust description and filename if necessary.
(sometimes
this page displays wrongly in my browser) [most of this step seems redundant, as you can change all the information except the filename on
the
next page anyway] 6. Click Continue 7. Click Save Page 8. Click "Flickr Uploader" link. 9. Click "Comment" link 10. Type a comment, usually referring to the page you'll be adding the
image
to. Tense can be complicated as the image isn't actually visible *yet*. 11. Close the tab and return to your main tab. 12. Click the Edit link next to the original article, opening in a new
tab.
- Type the name of the image somewhere, either as part of an infobox,
or
an [[image:...]] link. Usually involves flipping back somewhere to get
the
exact image name. 14. Type an edit summary (usually "Add image") 15. Click save. 16. Check that everything looks ok. 17. Close the tab.
So, just to clarify, steps 4-17 are actually the upload tool :-)
Yup. I'm making observations about the whole process, not just your bit :)
Hmm. Categories on wikipedia are about articles, categories on commons
are about images. Even if we could somehow link the two, most images in the commons category would only apply to few articles in the wikipedia category.
"About images"? I don't quite understand - they're generally concepts, the way that categories in a Wikipedia are.
What's the difference between [[:en:Category:Martin Luther King, Jr]] and [[:Commons:Martin Luther King]]? Both refer to the same person. One has articles in that category, and the other has images, but the central concept - a very specific person - is the same.
So what I'm thinking is that if there was some sort of central repository of concepts, you could do something like this:
1. On the article in each Wikipedia, you stamp the article with a link to this concept. Say [[rep:Martin Luther King, Jr]]. 2. All the cross language links are automatically generated and maintained - everything stamped with that concept automatically links to every other language article that has that concept. 3. On the relevant article or category on Commons, you again link to the concept. 4. By default (or perhaps with a tag like {{{SHOWCOMMONSIMAGES|3}}}) images from the linked category are shown.
There are of course issues when different Wikipedias divide the world up different ways, and of course naming: are the concepts all in English, multiple languages (so that one wikipedia could link to "Turin" and another to "Torino" and they would both "find each other", or in some conventional local language (ie, the local transliteration of a name is always used). But in many cases, like names of people or places, you see this kind of thing a lot:
[[de:Henri Oreiller]] [[fi:Henri Oreiller]] [[fr:Henri Oreiller]] [[it:Henri Oreiller]] [[no:Henri Oreiller]] [[pl:Henri Oreiller]]
Now, when I translated that from French to English, theoretically I should have updated all the other interwiki links as well. But that's a lot of work. Bots help a bit, but a central repository would be "the next level".
Note that it might be easy to semi-automatically link categories
between wikipedias:
- Get all articles in a category on xx.wikipedia
- For each of these, find the language link to yy.wikipedia, and list
the categories of that article
- sort categories at yy.wikipedia by number of occurrence
- the most occurring category is most likely to be the same as the
original one on xx.wikipedia If this is desparately wanted, I might write Yet Another Tool ;-)
Sounds a bit hit-and-miss. I think we should be moving in the direction of attempting to actually encode higher levels of meta-information, rather than attempting to deduce such information by heuristics.
So, which category to chose for [[en:Horse-ripping]]?
Ah, I think you've misunderstood. I'm suggesting using a direct link to Commons, so in this case, either [[commons:Horse-ripping]] or [[Commons:Category:Horse-ripping]]. Not using the article's category.
- With a single click, the flickr image is prepared for upload to Commons
with appropriate categories. You simply tweak image name and
description,
then press save.
So the commons categories that match the (say) en categories of the article for which flickr images were found are added to the uploaded flickr image? Sounds good. But isn't that what CommonSense [1] does?
Maybe. CommonSense has never really worked for me, and it's not integrated fully. These tools have to move from being an optional third-party utility used by a tiny fraction of editors to being a fundamental part of the editing process.
- The Wikipedia page (or pages) is then shown with the new image for
approval. Since this is automatic, you simply have to check that there aren't any special issues to know about.
That would be the end point of any (manual or automated) upload process anyway, right?
I should have been clearer: I really do mean the Wikipedia article, no the Commons image page. So it would (in this hypothetical, perfect tool) edit the article text to add the new image for you.
Counter-suggestion: Have a commons upload tool that will
- Take the flickr URL and optionally a new filename
- Check if the flickr image has a suitable license
- Run CommonSense (or the like) to find appropriate categories on commons
- Upload the image to commons under the appropriate bot name, using
the new filename, flickr template, and categories 5. Open the final image page and present it to the user (the other steps are "invisible")
This is good. If MediaWiki supported image renaming and user-deletion, even better would be the following: 1. Take the flickr URL 2. Upload it and show the user the final, uploaded image with template and some guessed categories. 3. (Optional) User renames image, tweaks text, categories etc.
Alternatively, if my "upload from URL" extension were enabled on
commons, it could fill in the entire upload form. The user could check the wikitext before uploading, and upload under his own name with a single click.
I think the "upload under his own name" concept is problematic when you're uploading images that aren't self-made.
Steve
On 10/5/07, Steve Bennett stevagewp@gmail.com wrote:
"About images"? I don't quite understand - they're generally concepts, the way that categories in a Wikipedia are.
I think he meant "populated by images". Not a big deal.
[[de:Henri Oreiller]] [[fi:Henri Oreiller]] [[fr:Henri Oreiller]] [[it:Henri Oreiller]] [[no:Henri Oreiller]] [[pl:Henri Oreiller]]
Now, when I translated that from French to English, theoretically I should have updated all the other interwiki links as well. But that's a lot of work. Bots help a bit, but a central repository would be "the next level".
I agree that it would be nice if this exact same list existed in some "language neutral" location and you could just add [[en:Henri Oreiller]] to the bottom of it, which would, in theory, trigger some sort of internal process to purge the cache of all seven articles so that they infallibly show six interwiki links apiece after a moment's time. This would be even more helpful for topics where the number of languages has just increased from say, 59 to 60.
Sounds a bit hit-and-miss. I think we should be moving in the direction of attempting to actually encode higher levels of meta-information, rather than attempting to deduce such information by heuristics.
There may, of course, not always be a quid pro quo for whatever reason.
1. Difference in growth curve from project to project. This will resolve itself in due time. Not a big deal. 2. Semantics: speakers of language A may draw a clear distinction between the concepts denoted by terms AX and AY, which may translate to language B as BX and BY, but be used interchangeably in the latter language. 3. For some topics, the English Wikipedia has a "salted" page or a protected redirect while other languages have a proper article.
Ah, I think you've misunderstood. I'm suggesting using a direct link to Commons, so in this case, either [[commons:Horse-ripping]] or [[Commons:Category:Horse-ripping]]. Not using the article's category.
I sense that this is the new [[meta-syntactic variable]] meme for this mailing list. Lovely.
I should have been clearer: I really do mean the Wikipedia article, no the Commons image page. So it would (in this hypothetical, perfect tool) edit the article text to add the new image for you.
If placeholder images were more widely used and/or if infobox syntax was more standardized, this would be easy to do. Failing that there's always [[Image:$1|thumb|{{subst:PAGENAME}}]], which I put somewhere in the article when adding a commons image to an article on a project whose language I don't speak a word of. Somebody else can shuffle things around later. The important thing is that editors of the article know that the image is available to them.
There is of course the {{commons}} template in every language which links to an equivalently titled page on the commons. For topics where a different title is used in every language, that means somebody on commons ought to create a buttload of redirects (inevitably pointing to La Lingua Anglofonica) to ensure that everyone (literally) ends up on the same page.
Of course if there was a way to work the commons page into the shared interwiki list (stored, once again, somewhere in no man's land) this too could be made easier.
Alternatively, if my "upload from URL" extension were enabled on commons, it could fill in the entire upload form. The user could check the wikitext before uploading, and upload under his own name with a single click.
I think the "upload under his own name" concept is problematic when you're uploading images that aren't self-made.
I think he means "under his own user account". Not a big deal.
—C.W.
On 10/8/07, Charlotte Webb charlottethewebb@gmail.com wrote:
I agree that it would be nice if this exact same list existed in some "language neutral" location and you could just add [[en:Henri Oreiller]] to the bottom of it, which would, in theory, trigger some sort of internal process to purge the cache of all seven articles so that they infallibly show six interwiki links apiece after a moment's time. This would be even more helpful for topics where the number of languages has just increased from say, 59 to 60.
Yep.
2. Semantics: speakers of language A may draw a clear distinction
between the concepts denoted by terms AX and AY, which may translate to language B as BX and BY, but be used interchangeably in the latter language.
Yes, I alluded to that possibility. But towns and people, in particular, strike me as pretty clear cut. It's unusual to see a person only form part of an article rather than the whole thing. Though it does happen, such as with minor candidates for minor elections.
3. For some topics, the English Wikipedia has a "salted" page or a
protected redirect while other languages have a proper article.
Yes, and our hypothetical semantic linking system could handled that, I would think.
If placeholder images were more widely used and/or if infobox syntax was more standardized, this would be easy to do. Failing that there's always [[Image:$1|thumb|{{subst:PAGENAME}}]], which I put somewhere in
Hmm, I don't even bother with a caption. But each to his own.
Of course if there was a way to work the commons page into the shared interwiki list (stored, once again, somewhere in no man's land) this too could be made easier.
In fact, if there was a way to transclude from another wiki, the problem is solved I think.
{{Commons:iw:Horse-ripping}} could simply return a list of all the interwiki links for that topic. We might need to be slightly smarter about handling interwiki self links (ie, from en to en).
Steve
On 10/7/07, Steve Bennett stevagewp@gmail.com wrote:
- Semantics: speakers of language A may draw a clear distinction
between the concepts denoted by terms AX and AY, which may translate to language B as BX and BY, but be used interchangeably in the latter language.
Yes, I alluded to that possibility. But towns and people, in particular, strike me as pretty clear cut. It's unusual to see a person only form part of an article rather than the whole thing. Though it does happen, such as with minor candidates for minor elections.
I was thinking more of abstract terms, which may for whatever cultural reasons, be difficult to properly match.
In fact, if there was a way to transclude from another wiki, the problem is solved I think.
{{Commons:iw:Horse-ripping}} could simply return a list of all the interwiki links for that topic. We might need to be slightly smarter about handling interwiki self links (ie, from en to en).
If this is to be a cross-wiki transclusion, as an ungraceful hack, something like <div style="<includeonly>display:none;</includeonly>"> wrapped around the interwiki list would prevent the self-link from appearing (bold-faced) at the bottom of the article area.
—C.W.
On 10/8/07, Charlotte Webb charlottethewebb@gmail.com wrote:
I was thinking more of abstract terms, which may for whatever cultural reasons, be difficult to properly match.
Yes, they exist, and there are cases that are probably intractable. That's why I'm sticking to the easy problems :) A solution which covers 30-50% of articles would still be very beneficial. It doesn't have to do everything.
If this is to be a cross-wiki transclusion, as an ungraceful hack,
something like <div style="<includeonly>display:none;</includeonly>"> wrapped around the interwiki list would prevent the self-link from appearing (bold-faced) at the bottom of the article area.
Really? I'll have to take your word for it.
Steve
On 10/8/07, Charlotte Webb charlottethewebb@gmail.com wrote:
If this is to be a cross-wiki transclusion, as an ungraceful hack, something like <div style="<includeonly>display:none;</includeonly>"> wrapped around the interwiki list would prevent the self-link from appearing (bold-faced) at the bottom of the article area.
Interestingly, I just came across a commons link included as if it were a normal interwiki language link:
http://sk.wikipedia.org/wiki/Hrebienok
I guess this is against policy, but it's kind of cool.
When you think about it, it would actually make more sense for our links to Wiktionary, Commons, Wikiquote etc to be displayed in the same way as links to other languages.
Steve
On 10/7/07, Steve Bennett stevagewp@gmail.com wrote:
Interestingly, I just came across a commons link included as if it were a normal interwiki language link:
http://sk.wikipedia.org/wiki/Hrebienok
I guess this is against policy, but it's kind of cool.
When you think about it, it would actually make more sense for our links to Wiktionary, Commons, Wikiquote etc to be displayed in the same way as links to other languages.
I don't know if there's any policy or just nobody here thought of it.
Apparently the trick is that the template they use on the Slovakian Wikipedia for project links[1] contains <div id="interProject"> and their monobook [2] contains a javascript hack that puts such things in the side-bar. Apparently this feature was copied from the Italian Wikipedia.
[1] http://sk.wikipedia.org/wiki/%C5%A0abl%C3%B3na:Projekt/odkaz [2] http://sk.wikipedia.org/wiki/MediaWiki:Monobook.js
—C.W.
More suggestions: if the general goal is to find images to add to articles that don't have any, then there are two obvious places to also look: - On Commons, try looking for an article or category matching the article name - On other Wikipedias, follow the interwiki links. You may end up on Commons anyway.
As an example, I just added an image for [[Tatranska Lomnica]]. But then it turned out there is a *massive* collection of related images on Commons: http://commons.wikimedia.org/wiki/Category:Tatransk%C3%A1_Lomnica
Some convenient way of displaying these images (or at least the count of them) would help.
Steve
I thought you could just type whatever image it is (eg. [[image:image from commons.jpg]]) on any of our projects and have the image appear. Maybe I'm wrong
Phoenix 15
On 10/7/07, Phoenix wiki phoenix.wiki@gmail.com wrote:
I thought you could just type whatever image it is (eg. [[image:image from commons.jpg]]) on any of our projects and have the image appear. Maybe I'm wrong
You are correct except for in cases where a different image exists at the same title on the "local" project. Unfortunately the software does not attempt to prevent this from happening accidentally (no warning to the latter uploader), as far as I know.
—C.W.
Charlotte Webb wrote:
On 10/7/07, Phoenix wiki phoenix.wiki@gmail.com wrote:
I thought you could just type whatever image it is (eg. [[image:image from commons.jpg]]) on any of our projects and have the image appear. Maybe I'm wrong
You are correct except for in cases where a different image exists at the same title on the "local" project. Unfortunately the software does not attempt to prevent this from happening accidentally (no warning to the latter uploader), as far as I know.
I believe it does prevent an image from being uploaded locally when an image with the same name is on Commons. Note that this doesn't work when the local copy is first.
On 10/7/07, Steve Bennett stevagewp@gmail.com wrote:
More suggestions: if the general goal is to find images to add to articles that don't have any, then there are two obvious places to also look:
- On Commons, try looking for an article or category matching the article
name
That would be a site-specific google-search, then? The API search apparently doesn't work, and exact matching strikes me as too slim a chance. For example, that would not have worked for your example, [[Tatranska Lomnica]].
- On other Wikipedias, follow the interwiki links. You may end up on Commons
anyway.
Apparently, you haven't seen my "Missing images" tool: http://tools.wikimedia.de/~magnus/missing_images.php (still working on the CSV mode :-)
Cheers, Magnus
On 10/7/07, Magnus Manske magnusmanske@googlemail.com wrote:
That would be a site-specific google-search, then? The API search
No, I'm just suggesting looking up the specific name.
apparently doesn't work, and exact matching strikes me as too slim a
chance. For example, that would not have worked for your example, [[Tatranska Lomnica]].
Actually, it would, I didn't spell it precisely. There's an accent on the third a - see where [[en:...]] redirects to.
- On other Wikipedias, follow the interwiki links. You may end up on
Commons
anyway.
Apparently, you haven't seen my "Missing images" tool: http://tools.wikimedia.de/~magnus/missing_images.php (still working on the CSV mode :-)
Oh, I don't doubt it. I'm suggesting integrating these "tools" to create "software" :) 5 different tools is good. One piece of great software that does all that is even better.
So, basically, I'm thinking of software that attempts to solve the general problem of "this article has no images. Whence can we procure some?" Possible sources include commons, interwikis, flickr, and potentially dozens of other sources.
Steve