---------- Forwarded message ---------- From: Olli ollinpostit@gmail.com Date: 2009/10/31 Subject: Wikipedia christmas calendar? To: translators-l@lists.wikimedia.org
What about a wikipedia christmas calendar? It can maybe preview some articles or something similar. Then it can be multilingual.
What do you think?
Olli
*(am I using wrong mailing list?)*
Olli wrote:
Date: 2009/10/31 Subject: Wikipedia christmas calendar?
What about a wikipedia christmas calendar? It can maybe preview some articles or something similar. Then it can be multilingual.
Not necessarily just Christmas, but a published calendar for the whole year. Wikipedia's date articles are already full of weitd and wonderful things that happened on this day in history.
Ec
On Sat, Oct 31, 2009 at 5:09 PM, Ray Saintonge saintonge@telus.net wrote:
Olli wrote:
Date: 2009/10/31 Subject: Wikipedia christmas calendar?
What about a wikipedia christmas calendar? It can maybe preview some articles or something similar. Then it can be multilingual.
Not necessarily just Christmas, but a published calendar for the whole year. Wikipedia's date articles are already full of weitd and wonderful things that happened on this day in history.
Or, have a new calender generated automatically on each page reload:
http://toolserver.org/~magnus/wikilendar.php?month=11&year=2009
(takes a few seconds to parse all these pages and check for suitable images)
Cheers, Magnus
2009/11/2 Magnus Manske magnusmanske@googlemail.com:
On Sat, Oct 31, 2009 at 5:09 PM, Ray Saintonge saintonge@telus.net wrote:
Olli wrote:
Date: 2009/10/31 Subject: Wikipedia christmas calendar?
What about a wikipedia christmas calendar? It can maybe preview some articles or something similar. Then it can be multilingual.
Not necessarily just Christmas, but a published calendar for the whole year. Wikipedia's date articles are already full of weitd and wonderful things that happened on this day in history.
Or, have a new calender generated automatically on each page reload:
http://toolserver.org/~magnus/wikilendar.php?month=11&year=2009
(takes a few seconds to parse all these pages and check for suitable images)
I think a human selected calendar would be better for actually publishing, but that's certainly a fun script. Can you add an option to choose a category so we can have a calendar of Britons, or French, or mathematicians, or military people, etc.?
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
On Mon, Nov 2, 2009 at 4:54 PM, Thomas Dalton thomas.dalton@gmail.com wrote:
2009/11/2 Magnus Manske magnusmanske@googlemail.com:
On Sat, Oct 31, 2009 at 5:09 PM, Ray Saintonge saintonge@telus.net wrote:
Olli wrote:
Date: 2009/10/31 Subject: Wikipedia christmas calendar?
What about a wikipedia christmas calendar? It can maybe preview some articles or something similar. Then it can be multilingual.
Not necessarily just Christmas, but a published calendar for the whole year. Wikipedia's date articles are already full of weitd and wonderful things that happened on this day in history.
Or, have a new calender generated automatically on each page reload:
http://toolserver.org/~magnus/wikilendar.php?month=11&year=2009
(takes a few seconds to parse all these pages and check for suitable images)
I think a human selected calendar would be better for actually publishing, but that's certainly a fun script. Can you add an option to choose a category so we can have a calendar of Britons, or French, or mathematicians, or military people, etc.?
I could limit the articles used to a ceratin category, but IMHO that would restrict the search too much, that is, either no or a few possible candidates per day.
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article from e.g. [[November 2]], then chose a random picture from that.
Cheers, Magnus
2009/11/2 Magnus Manske magnusmanske@googlemail.com:
I think a human selected calendar would be better for actually publishing, but that's certainly a fun script. Can you add an option to choose a category so we can have a calendar of Britons, or French, or mathematicians, or military people, etc.?
I could limit the articles used to a ceratin category, but IMHO that would restrict the search too much, that is, either no or a few possible candidates per day.
For large categories, like nationalities, it would probably work. For smaller categories you would need to decide what to do with gaps.
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article from e.g. [[November 2]], then chose a random picture from that.
Try the image in the infobox if there is one.
On Mon, Nov 2, 2009 at 5:03 PM, Thomas Dalton thomas.dalton@gmail.com wrote:
2009/11/2 Magnus Manske magnusmanske@googlemail.com:
I think a human selected calendar would be better for actually publishing, but that's certainly a fun script. Can you add an option to choose a category so we can have a calendar of Britons, or French, or mathematicians, or military people, etc.?
I could limit the articles used to a ceratin category, but IMHO that would restrict the search too much, that is, either no or a few possible candidates per day.
For large categories, like nationalities, it would probably work. For smaller categories you would need to decide what to do with gaps.
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article from e.g. [[November 2]], then chose a random picture from that.
Try the image in the infobox if there is one.
So, either a HTML parser or a new MediaWiki parser to get at the <img>/variable. Tough choice there :-)
2009/11/2 Magnus Manske magnusmanske@googlemail.com:
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article from e.g. [[November 2]], then chose a random picture from that.
First image is probably your best bet - the odds are reasonably high it'll be a picture, or something else "representative", in the conventional top-right slot. Certainly better odds than random selection!
suggestions:
1 =You should add option to select what country (what wikipedia (fi, en, sv...)) 2 = Calendr should be like right calendar 3 = Other languages (translation possiblity)
Olli
2009/11/2 Andrew Gray andrew.gray@dunelm.org.uk
2009/11/2 Magnus Manske magnusmanske@googlemail.com:
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article from e.g. [[November 2]], then chose a random picture from that.
First image is probably your best bet - the odds are reasonably high it'll be a picture, or something else "representative", in the conventional top-right slot. Certainly better odds than random selection!
--
- Andrew Gray
andrew.gray@dunelm.org.uk
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
Дана Monday 02 November 2009 18:31:50 Andrew Gray написа:
First image is probably your best bet - the odds are reasonably high it'll be a picture, or something else "representative", in the conventional top-right slot. Certainly better odds than random selection!
First image could easily be an icon.
2009/11/2 Nikola Smolenski smolensk@eunet.rs:
Дана Monday 02 November 2009 18:31:50 Andrew Gray написа:
First image is probably your best bet - the odds are reasonably high it'll be a picture, or something else "representative", in the conventional top-right slot. Certainly better odds than random selection!
First image could easily be an icon.
How about first image more than 50px (or whatever) wide?
2009/11/2 Nikola Smolenski smolensk@eunet.rs:
Дана Monday 02 November 2009 18:31:50 Andrew Gray написа:
First image is probably your best bet - the odds are reasonably high it'll be a picture, or something else "representative", in the conventional top-right slot. Certainly better odds than random selection!
First image could easily be an icon.
Mmm, true. "First image directly invoked in the wikitext", or something, I guess.
I'm not sure how the standard popups select their image, but that seems a fairly efficient method to look at - it fails sometimes, but in the cases where it fails there's rarely a better image it *should* have used.
On Tue, Nov 3, 2009 at 6:53 PM, Andrew Gray andrew.gray@dunelm.org.uk wrote:
2009/11/2 Nikola Smolenski smolensk@eunet.rs:
Дана Monday 02 November 2009 18:31:50 Andrew Gray написа:
First image is probably your best bet - the odds are reasonably high it'll be a picture, or something else "representative", in the conventional top-right slot. Certainly better odds than random selection!
First image could easily be an icon.
Mmm, true. "First image directly invoked in the wikitext", or something, I guess.
So, no images from infoboxes, then. (Except those few that actually require [[File:..]]).
Cheers, Magnus
Magnus Manske wrote:
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article
Google image search has it, so it's not that complex. Not that I suggest you should implement it in PHP :)
Nikola Smolenski wrote:
Magnus Manske wrote:
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article
Google image search has it, so it's not that complex. Not that I suggest you should implement it in PHP :)
It *is* complex. Yes, it's not impossible; yes, it's been done. But that doesn't mean it's simple or easy. Btw, Magnus, thanks for building that demo. It's great for putting this idea in perspective! :)
Waldir Pimenta wrote:
Nikola Smolenski wrote:
Magnus Manske wrote:
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article
Google image search has it, so it's not that complex. Not that I suggest you should implement it in PHP :)
It *is* complex. Yes, it's not impossible; yes, it's been done. But that doesn't mean it's simple or easy. Btw, Magnus, thanks for building that demo. It's great for putting this idea in perspective! :) _______________________________________________ foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
There's a review of the state of the art in face detection at http://vision.ai.uiuc.edu/mhyang/face-detection-survey.html , if you are looking for ideas for algorithms.
-- Neil
On Tue, Nov 3, 2009 at 2:51 AM, Waldir Pimenta waldir@email.com wrote:
Nikola Smolenski wrote:
Magnus Manske wrote:
One bug: I got a graph of Imran Khan's bowling statistics rather than his portrait...
And if you give me code to identify a person's image, I'll be happy to implement it, as would the NSA. As it stands, I chose a random article
Google image search has it, so it's not that complex. Not that I suggest you should implement it in PHP :)
It *is* complex. Yes, it's not impossible; yes, it's been done. But that doesn't mean it's simple or easy.
Are you talking about face detection or facial recognition? The former is implemented in some cell phones. It's "not that complex". The latter is more cutting edge, though surely the NSA has a good implementation of it (as do many casinos).
Selecting pictures of people as opposed to bowling statistics would be the former, right?
On Tue, Nov 3, 2009 at 2:19 PM, Anthony wikimail@inbox.org wrote:
Are you talking about face detection or facial recognition? The former is implemented in some cell phones. It's "not that complex". The latter is more cutting edge, though surely the NSA has a good implementation of it (as do many casinos).
Selecting pictures of people as opposed to bowling statistics would be the former, right?
OK, so shouldn't we run this against all of Commons, then? If we could offer "show me only pictures of people" as a checkbox in search, that would surely be appreciated (and useful for the calendar thingy as well;-)
Cheers, Magnus
Magnus Manske wrote:
OK, so shouldn't we run this against all of Commons, then? If we could offer "show me only pictures of people" as a checkbox in search, that would surely be appreciated (and useful for the calendar thingy as well;-)
The link that Neil posted led me to http://opencv.willowgarage.com/wiki/FaceDetection and that in turn led me to http://www.xarg.org/project/php-facedetect/ - this isn't as impossible as it seemed to be a few minutes ago! :)
On Tue, Nov 3, 2009 at 9:37 AM, Magnus Manske magnusmanske@googlemail.com wrote:
On Tue, Nov 3, 2009 at 2:19 PM, Anthony wikimail@inbox.org wrote:
Are you talking about face detection or facial recognition? The former is implemented in some cell phones. It's "not that complex". The latter is more cutting edge, though surely the NSA has a good implementation of it (as do many casinos).
Selecting pictures of people as opposed to bowling statistics would be the former, right?
OK, so shouldn't we run this against all of Commons, then?
As long as you don't set up a bot to add categories without checking the results manually, sure. IMO any automated results are better kept in a specialized database because they'll likely be changing rapidly. Maybe on the toolserver.
2009/10/31 Olli ollinpostit@gmail.com:
---------- Forwarded message ---------- From: Olli ollinpostit@gmail.com Date: 2009/10/31 Subject: Wikipedia christmas calendar? To: translators-l@lists.wikimedia.org
What about a wikipedia christmas calendar? It can maybe preview some articles or something similar. Then it can be multilingual.
Last year we have created such a calendar:
http://commons.wikimedia.org/wiki/File:Kalendarz_Wikimedia_2009.pdf
It uses Common's pictures of the month + some Wikipedia related events + Polish Wikipedian's nicknames placed in the days where they have their birthday :-)
wikimedia-l@lists.wikimedia.org