I'd like to propose a new Wikipedia: "the multilingual wikipedia"
Some ideas about how to do that are on http://meta.wikimedia.org/wiki/Multilingual_communication
I would be very interested in getting the multilingual features of oddmuse (the colored background, the language filters) ported to Mediawiki and then start working on two things:
A) Elaborate the engine to make it work for multilingual pages: language filters also in the edit mode, inclusion of automatic pretranslation in a new translate-mode, etc. B) Start making the "multilingual wikipedia". Kinda getting people into the fun of translation.
Hopefully this wouldn't require a fork of Mediawiki. Unfortunately I have a vage feeling that it does. In that case I propose as a name for the multilingual Mediawiki engine: MMediawiki ;)
But who knows? Maybe things are easier than I think?
thank you
MattisManzel
On the topic of multilingual coordination at large, I have often thought when adding interwiki links to pages that it would be nice to have a central, language-independent repository of cross-language links.
(Probably these issues have been raised before; sorry if I'm rehashing old points or raising them in the wrong place.)
I've seen essentially two problems as is: - links to new translations don't get added everywherewikipedia - erroneous cross-language links can persist even after being corrected somewhere
Robots can solve the first problem but not the second. An example of the second is 'Germanic languages'. There are not too many wikipedias that have any entry for this topic specifically. However, someone had linked a number of them to the equivalent for 'German language' in other wikipedias, probably not realizing the difference.
I don't see how a robot could solve this problem without semantic knowledge of the equivalents of 'Germanic language' in various languages.
Steve
On Wed, 3 Nov 2004 12:56:51 -0500, Stephen Forrest stephen.forrest@gmail.com wrote:
I've seen essentially two problems as is:
- links to new translations don't get added everywherewikipedia
- erroneous cross-language links can persist even after being
corrected somewhere
Robots can solve the first problem but not the second. An example of the second is 'Germanic languages'. There are not too many wikipedias that have any entry for this topic specifically. However, someone had linked a number of them to the equivalent for 'German language' in other wikipedias, probably not realizing the difference.
I don't see how a robot could solve this problem without semantic knowledge of the equivalents of 'Germanic language' in various languages.
As a bot writer and operator, I can say that bots _could_ help in finding this kind of cases. We might go work more on it (the bot notices it if it can get to two pages on the same language - in this case "Germanic languages" and "German language"). There's quite a number of such problems; a common cause is that when a page turns into a disambiguation page on some Wikipedia, the interwiki links to it are usually not changed.
Andre Engels
On Wed, 3 Nov 2004 20:11:36 +0100, Andre Engels andreengels@gmail.com wrote:
As a bot writer and operator, I can say that bots _could_ help in finding this kind of cases. We might go work more on it (the bot notices it if it can get to two pages on the same language - in this case "Germanic languages" and "German language"). There's quite a number of such problems; a common cause is that when a page turns into a disambiguation page on some Wikipedia, the interwiki links to it are usually not changed.
I see your point that bots can help; my point was that they can't do so automatically. A bot would have to inform a human that both 'Germanic language' and 'German language' were encountered, and the human would have to take action based on this.
What I suggest is to have a language-independent 'linker' page maintained in a separate namespace. (Maybe 'meta' or something new, e.g. 'interwiki'). All it would consist of is a list of interwiki links, with the meaning that any page named on the linker page was to be taken to be an equivalent for any other named page.
Under this new scheme, suppose I'm reading the English wikipedia article on 'Germanic languages''. If I then notice a link to [[da:Tysk sprog]] and fix it, I will be editing this linker page. If the previous error had been propogated into other wikipedias, they will have their link corrected immediately, without having to wait for a bot or human to spot the bad link and fix it.
Some more advantages:
1) Since interwiki changes would then be changes to the linker page, they would not add volume to the changelist history of linked pages.
2) There would be fewer anonymous edits to wikipedias. Right now, if I want to add an interwiki link to a language I don't speak, I won't bother creating an account, because I wouldn't particularly benefit from having one. In this proposed system, I would merely need to have an account on 'meta' or 'interwiki'.
3) The bloat factor of interwiki links on pages would be reduced.
4) The problems with broken interwiki links caused by a wikipedia munging foreign characters entered as literals in the input would be more easily fixed. I've seen a few interwiki links to Cyrillic wikipedias appear as, for example, [[bg:????????]], probably because someone tried to copy-paste a literal character from one page to another, and their system didn't support the character set.
Hmm, that was a bit longer than I intended. :) Again, please feel free to give me feedback on the above idea, or redirect me to the appropriate forum to post it if this isn't it.
regards,
Steve
I'm not 100% sure, Steve, but I think this may have been proposed before.
Unfortunately everybody's very busy ATM, and an earlier proposal to share all media between all projects instead of having to upload it separately, Wikimedia Commons, was just launched recently I believe (that was another idea that lots of people had to improve things).
However that is definitely something we should pursue.
mark
On Wed, 3 Nov 2004 17:48:22 -0500, Stephen Forrest stephen.forrest@gmail.com wrote:
On Wed, 3 Nov 2004 20:11:36 +0100, Andre Engels andreengels@gmail.com wrote:
As a bot writer and operator, I can say that bots _could_ help in finding this kind of cases. We might go work more on it (the bot notices it if it can get to two pages on the same language - in this case "Germanic languages" and "German language"). There's quite a number of such problems; a common cause is that when a page turns into a disambiguation page on some Wikipedia, the interwiki links to it are usually not changed.
I see your point that bots can help; my point was that they can't do so automatically. A bot would have to inform a human that both 'Germanic language' and 'German language' were encountered, and the human would have to take action based on this.
What I suggest is to have a language-independent 'linker' page maintained in a separate namespace. (Maybe 'meta' or something new, e.g. 'interwiki'). All it would consist of is a list of interwiki links, with the meaning that any page named on the linker page was to be taken to be an equivalent for any other named page.
Under this new scheme, suppose I'm reading the English wikipedia article on 'Germanic languages''. If I then notice a link to [[da:Tysk sprog]] and fix it, I will be editing this linker page. If the previous error had been propogated into other wikipedias, they will have their link corrected immediately, without having to wait for a bot or human to spot the bad link and fix it.
Some more advantages:
- Since interwiki changes would then be changes to the linker page,
they would not add volume to the changelist history of linked pages.
- There would be fewer anonymous edits to wikipedias. Right now, if
I want to add an interwiki link to a language I don't speak, I won't bother creating an account, because I wouldn't particularly benefit from having one. In this proposed system, I would merely need to have an account on 'meta' or 'interwiki'.
The bloat factor of interwiki links on pages would be reduced.
The problems with broken interwiki links caused by a wikipedia
munging foreign characters entered as literals in the input would be more easily fixed. I've seen a few interwiki links to Cyrillic wikipedias appear as, for example, [[bg:????????]], probably because someone tried to copy-paste a literal character from one page to another, and their system didn't support the character set.
Hmm, that was a bit longer than I intended. :) Again, please feel free to give me feedback on the above idea, or redirect me to the appropriate forum to post it if this isn't it.
regards,
Steve
Wikipedia-l mailing list Wikipedia-l@Wikimedia.org http://mail.wikipedia.org/mailman/listinfo/wikipedia-l
Ah yes, if what you meant to say is that this system is better, I definitely agree with you. It's an idea that has been floating around more. Only problem is the amount of coding it would take, and that one would need to get information from more than one database to show a single page.
Andre Engels
On Wed, 3 Nov 2004 17:48:22 -0500, Stephen Forrest stephen.forrest@gmail.com wrote:
On Wed, 3 Nov 2004 20:11:36 +0100, Andre Engels andreengels@gmail.com wrote:
As a bot writer and operator, I can say that bots _could_ help in finding this kind of cases. We might go work more on it (the bot notices it if it can get to two pages on the same language - in this case "Germanic languages" and "German language"). There's quite a number of such problems; a common cause is that when a page turns into a disambiguation page on some Wikipedia, the interwiki links to it are usually not changed.
I see your point that bots can help; my point was that they can't do so automatically. A bot would have to inform a human that both 'Germanic language' and 'German language' were encountered, and the human would have to take action based on this.
What I suggest is to have a language-independent 'linker' page maintained in a separate namespace. (Maybe 'meta' or something new, e.g. 'interwiki'). All it would consist of is a list of interwiki links, with the meaning that any page named on the linker page was to be taken to be an equivalent for any other named page.
Under this new scheme, suppose I'm reading the English wikipedia article on 'Germanic languages''. If I then notice a link to [[da:Tysk sprog]] and fix it, I will be editing this linker page. If the previous error had been propogated into other wikipedias, they will have their link corrected immediately, without having to wait for a bot or human to spot the bad link and fix it.
Some more advantages:
- Since interwiki changes would then be changes to the linker page,
they would not add volume to the changelist history of linked pages.
- There would be fewer anonymous edits to wikipedias. Right now, if
I want to add an interwiki link to a language I don't speak, I won't bother creating an account, because I wouldn't particularly benefit from having one. In this proposed system, I would merely need to have an account on 'meta' or 'interwiki'.
The bloat factor of interwiki links on pages would be reduced.
The problems with broken interwiki links caused by a wikipedia
munging foreign characters entered as literals in the input would be more easily fixed. I've seen a few interwiki links to Cyrillic wikipedias appear as, for example, [[bg:????????]], probably because someone tried to copy-paste a literal character from one page to another, and their system didn't support the character set.
Hmm, that was a bit longer than I intended. :) Again, please feel free to give me feedback on the above idea, or redirect me to the appropriate forum to post it if this isn't it.
regards,
Steve
One question worths investigating is whether there is any case when links are not desired to be two-way: Article A1 links to B2 but B2 doesn't want to link back to A1 (or wants to link to A2 instead). I don't know whether there is such a case, maybe someone of you knows one.
I believe we use en: as the planned interwiki repository: if a link gets into en: then bots propagate it around. Bad links in en: propagate, too. If there was a single repository of interlanguage links these errors could be fixed easily. Now it's pretty hard, even with scripts and bots; usually needs a natural unintelligence machine (a Human Being(Tm)).
grin
On Thu, 4 Nov 2004 12:19:21 +0100, Peter Gervai grin@tolna.net wrote:
One question worths investigating is whether there is any case when links are not desired to be two-way: Article A1 links to B2 but B2 doesn't want to link back to A1 (or wants to link to A2 instead). I don't know whether there is such a case, maybe someone of you knows one.
I could see someone making a case for linking to 'near-equivalents'. Using my previous example, if I'm interested in reading about Germanic languages in Danish, but no such article exists, I might be willing to settle for reading an article about the German language.
I don't like this idea. I think when one adds an interwiki link one should be implicitly saying that the linked page covers exactly the same subject as the current page (not necessarily with the same content). To use mathematical terminology, the act of interwiki-linking should be an equivalence relation.
Steve
On Wednesday 03 November 2004 19:08, Mattis Manzel wrote:
Hopefully this wouldn't require a fork of Mediawiki.
I am working (a bit slowly, but still working) on a redistribution of mediawiki1.3.7 with some new features. If you are interested you can join me and add the features you want.
NSK schrieb:
On Wednesday 03 November 2004 19:08, Mattis Manzel wrote:
Hopefully this wouldn't require a fork of Mediawiki.
I am working (a bit slowly, but still working) on a redistribution of mediawiki1.3.7 with some new features. If you are interested you can join me and add the features you want.
hi NSK,
where?
mattis
On Wednesday 03 November 2004 22:09, Mattis Manzel wrote:
where?
Here: http://maatworks.wikinerds.org/index.php/WikiAnt
Mattis Manzel wrote:
I'd like to propose a new Wikipedia: "the multilingual wikipedia"
Some ideas about how to do that are on http://meta.wikimedia.org/wiki/Multilingual_communication
I would be very interested in getting the multilingual features of oddmuse (the colored background, the language filters) ported to Mediawiki and then start working on two things:
A) Elaborate the engine to make it work for multilingual pages: language filters also in the edit mode, inclusion of automatic pretranslation in a new translate-mode, etc. B) Start making the "multilingual wikipedia". Kinda getting people into the fun of translation.
Hopefully this wouldn't require a fork of Mediawiki. Unfortunately I have a vage feeling that it does. In that case I propose as a name for the multilingual Mediawiki engine: MMediawiki ;)
But who knows? Maybe things are easier than I think?
Or more difficult. I think the idea is full of naïve optimism about translation in general, and machine translation in particular. Begin with an English text and translate it into another language. Have someone else, who cannot see the original, translate it back into English. Compare the two English versions. The difference can be amazing.
On another level, articles on the same subject are often separately developed in each language. This can often result in having stable NPOVs in each language that are different, frequently in the wake of careful negotiation over the exact meaning of a word. Reconciling these could be extremely difficult. I can easily imagine that attitudes toward US foreign policy might not be as conflicted when written in many languages other than English. En
You mean "ec" ;)
On Wed, 03 Nov 2004 11:53:50 -0800, Ray Saintonge saintonge@telus.net wrote:
Mattis Manzel wrote:
I'd like to propose a new Wikipedia: "the multilingual wikipedia"
Some ideas about how to do that are on http://meta.wikimedia.org/wiki/Multilingual_communication
I would be very interested in getting the multilingual features of oddmuse (the colored background, the language filters) ported to Mediawiki and then start working on two things:
A) Elaborate the engine to make it work for multilingual pages: language filters also in the edit mode, inclusion of automatic pretranslation in a new translate-mode, etc. B) Start making the "multilingual wikipedia". Kinda getting people into the fun of translation.
Hopefully this wouldn't require a fork of Mediawiki. Unfortunately I have a vage feeling that it does. In that case I propose as a name for the multilingual Mediawiki engine: MMediawiki ;)
But who knows? Maybe things are easier than I think?
Or more difficult. I think the idea is full of naïve optimism about translation in general, and machine translation in particular. Begin with an English text and translate it into another language. Have someone else, who cannot see the original, translate it back into English. Compare the two English versions. The difference can be amazing.
On another level, articles on the same subject are often separately developed in each language. This can often result in having stable NPOVs in each language that are different, frequently in the wake of careful negotiation over the exact meaning of a word. Reconciling these could be extremely difficult. I can easily imagine that attitudes toward US foreign policy might not be as conflicted when written in many languages other than English. En
Wikipedia-l mailing list Wikipedia-l@Wikimedia.org http://mail.wikipedia.org/mailman/listinfo/wikipedia-l
Ray Saintonge schrieb:
Mattis Manzel wrote:
I'd like to propose a new Wikipedia: "the multilingual wikipedia"
Some ideas about how to do that are on http://meta.wikimedia.org/wiki/Multilingual_communication
I would be very interested in getting the multilingual features of oddmuse (the colored background, the language filters) ported to Mediawiki and then start working on two things:
A) Elaborate the engine to make it work for multilingual pages: language filters also in the edit mode, inclusion of automatic pretranslation in a new translate-mode, etc. B) Start making the "multilingual wikipedia". Kinda getting people into the fun of translation.
Hopefully this wouldn't require a fork of Mediawiki. Unfortunately I have a vage feeling that it does. In that case I propose as a name for the multilingual Mediawiki engine: MMediawiki ;)
But who knows? Maybe things are easier than I think?
Or more difficult. I think the idea is full of naïve optimism about translation in general, and machine translation in particular. Begin with an English text and translate it into another language. Have someone else, who cannot see the original, translate it back into English. Compare the two English versions. The difference can be amazing.
On another level, articles on the same subject are often separately developed in each language. This can often result in having stable NPOVs in each language that are different, frequently in the wake of careful negotiation over the exact meaning of a word. Reconciling these could be extremely difficult. I can easily imagine that attitudes toward US foreign policy might not be as conflicted when written in many languages other than English. En
Wikipedia-l mailing list Wikipedia-l@Wikimedia.org http://mail.wikipedia.org/mailman/listinfo/wikipedia-l
So, is it multiple worlds or do we live in one world? Do the elections in the US concen us here in Europe or do they not? Does it makes sense to effecivate wiki by making it a global thing instead of the thing of a signle langage pool like English or German or not? When we can (try) so?
So does NPOV existt at all or is it but a crusaders dream? Multiple point of view. MPOV, Coexistance and collaboration. ;) In short: terms wiki.
Mattis Manzel wrote:
Ray Saintonge schrieb:
Mattis Manzel wrote:
I'd like to propose a new Wikipedia: "the multilingual wikipedia"
Some ideas about how to do that are on http://meta.wikimedia.org/wiki/Multilingual_communication
I would be very interested in getting the multilingual features of oddmuse (the colored background, the language filters) ported to Mediawiki and then start working on two things:
A) Elaborate the engine to make it work for multilingual pages: language filters also in the edit mode, inclusion of automatic pretranslation in a new translate-mode, etc. B) Start making the "multilingual wikipedia". Kinda getting people into the fun of translation.
Hopefully this wouldn't require a fork of Mediawiki. Unfortunately I have a vage feeling that it does. In that case I propose as a name for the multilingual Mediawiki engine: MMediawiki ;)
But who knows? Maybe things are easier than I think?
Or more difficult. I think the idea is full of naïve optimism about translation in general, and machine translation in particular. Begin with an English text and translate it into another language. Have someone else, who cannot see the original, translate it back into English. Compare the two English versions. The difference can be amazing.
On another level, articles on the same subject are often separately developed in each language. This can often result in having stable NPOVs in each language that are different, frequently in the wake of careful negotiation over the exact meaning of a word. Reconciling these could be extremely difficult. I can easily imagine that attitudes toward US foreign policy might not be as conflicted when written in many languages other than English. Ec
So, is it multiple worlds or do we live in one world? Do the elections in the US concen us here in Europe or do they not? Does it makes sense to effecivate wiki by making it a global thing instead of the thing of a signle langage pool like English or German or not? When we can (try) so?
So does NPOV existt at all or is it but a crusaders dream? Multiple point of view. MPOV, Coexistance and collaboration. ;) In short: terms wiki.
The NPOV principle remains. Our Muslim Wikipedians would never support a crusader's dream. ;-) . There may be an ideal of one overall NPOV but there are a lot of social and cultural bridges to cross before we get there. When we have terms like "German language" and "Germanic language" it is relatively easy to identify, define and fix the problem. It is quite a different thing when values are involved. In the past we have had ongoing debates about what to name the rivers on the boundary between Germany and Poland. Those who participate on the German and Polish Wikipedias can rapidly reach their own respective though different NPOV consensus. When a third language must be thrown into the mix the result there is not so clear.
Of course US elections are of concern to Europe. But Americans might well respond, "It's our elections, so mind your own business?" Those Americans, however, who would argue in favour of the divine right of their President George III may find themselves at a disadvantage when they must do so in a peasant language. There is much to be reconciled there, but doing so will require much more than a software solution.
Ec
wikipedia-l@lists.wikimedia.org