Re: [Wikitech-l] Interwiki recommendations
Lars Aronsson skrev: These are two recipies for how you can improve your language version of Wikipedia in an evening. But we don't have any cookbook to collect all such recipies, do we? Should we? These two recipies have now been used for the Swedish Wikipedia and we don't need to follow them again. But they can be reused on every other language. Hi Lars, It'd probably a good idea to start up a collection of Wiki Patterns somewhere, in English. With some kind of voting for the best ideas/practices. Then the best ideas might spread. Regards, // Rolf Lampa ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Re: [Wikitech-l] Interwiki recommendations
On Tue, Feb 3, 2009 at 12:26 PM, Gerard Meijssen gerard.meijs...@gmail.comwrote: Hoi, Have a look at this: http://www.omegawiki.org/Expression:Nederland This is structured data It can be shown in multiple languages. It does allow for interwiki links... It already works with MediaWiki ... So the question is, why re-invent the wheel ? Thanks, GerardM 2009/2/3 Michael Dale md...@wikimedia.org We really need a wikidata type site. We ran into similar issues with structured data between government data wikis. Yaron hacked up a (relatively simple) extension called External Data for pulling external data into a given page. This ends up working very well, allowing us to effortlessly transclude shared datasets into templates of multiple wikis. This is fundamentally good as it moves queriable maintained structured data away from multiple instances of user maintained semi-structured data. For the wikimedia context I think something like wikidata.wikimedia.org needs to be created. It could be a semanticMediaWiki wiki installation extended with localized page aliases. A single page-id or concept would have many title columns for each language. (The localized title columns can be propagated by the existing database of inter-wiki language links). Furthermore since properties/relations have titles/page-ids; they could also be localized. Allowing you to query the shared structured dataset in your local language. Then something like external data extension will tie wikidata to all the current language wikis. This can be thought of as commons but for data. (likewise external to wikimedia wikis could use this structured data). This lets template authors concentrate on localized representation of the data (calling the native language properties) , articles authors can focus on the article (instead of huge seed of hard to maintain template data), and structured data folks can focus on importing data into the central shared repository. --michael I think Michael Dale said the same thing. He was just saying that perhaps we could have a central wiki (based on SMW) to hold the data, and allow the individual projects to use ExternalData to grab the data into the presentation format of their choice. Not a terrible idea, IMO :) Combining enwiki and frwiki into one mega-pedia isn't the best idea. Each wiki should be free to not only decide content as they so choose, but also to use the presentation they prefer. However, there's no need to duplicate raw facts as mentioned, so a central wiki of the raw facts query-able by ExternalData would serve this purpose. Allow wikis to retain their autonomy, but give them access to centralized data that doesn't change in different languages. -Chad ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Re: [Wikitech-l] Interwiki recommendations
Gerard Meijssen hett schreven: Hoi, Have a look at this: http://www.omegawiki.org/Expression:Nederland This is structured data It can be shown in multiple languages. It does allow for interwiki links... It already works with MediaWiki ... So the question is, why re-invent the wheel ? Thanks, GerardM Is it transcludable to other wikis? Marcus Buck ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Re: [Wikitech-l] Interwiki recommendations
Gerard Meijssen schreef: Hoi, Have a look at this: http://www.omegawiki.org/Expression:Nederland This is structured data It can be shown in multiple languages. It does allow for interwiki links... It already works with MediaWiki ... So the question is, why re-invent the wheel ? I think the wheel is called Semantic MediaWiki in this case :P Roan Kattouw (Catrope) ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Re: [Wikitech-l] Interwiki recommendations
hmm ... my feeling is that it would be easier to adapt semantic wiki to include language aliasing than adapt omega wiki to be less dictionary centric in the context of a shared structured data site with _lossy_ defined community ontologies... But fundamentally there is no reason why an external data extension could not pull from both such external structured data systems. to elaborate a bit... As far as I can tell OmegaWiki would have to go a long way to be used in the same way that semantic wiki is used today. And an even further way to integrate with how wikimedia uses templates and infoboxes today. Omega seems to tie relations to items in a specif defined concepts name space, rather than arbitrary wiki pages. This makes a lot of sense if designing a multi-lingual language representation system but not so ideal for a general purpose shared structured data template propagation system tied to specif article entries. OmegaWiki is seems to be focused on multilengual language representation rather than multiple data type representation, re-usage, and template integration. Semantic wiki appears to be more of a flexible platform in this area as it has spawned dozens of extensions and hundreds of usages in a wide range of contexts. Semantic wiki has already done a few revisions to optimize the storage and retrieval of multiple data-types, its more closely tied to the svn version of mediaWiki, they do regular releases etc. That being said let me reiterate an external data extension could support multiple remote data systems. --michael Gerard Meijssen wrote: Hoi, Have a look at this: http://www.omegawiki.org/Expression:Nederland This is structured data It can be shown in multiple languages. It does allow for interwiki links... It already works with MediaWiki ... So the question is, why re-invent the wheel ? Thanks, GerardM 2009/2/3 Michael Dale md...@wikimedia.org We really need a wikidata type site. We ran into similar issues with structured data between government data wikis. Yaron hacked up a (relatively simple) extension called External Data for pulling external data into a given page. This ends up working very well, allowing us to effortlessly transclude shared datasets into templates of multiple wikis. This is fundamentally good as it moves queriable maintained structured data away from multiple instances of user maintained semi-structured data. For the wikimedia context I think something like wikidata.wikimedia.org needs to be created. It could be a semanticMediaWiki wiki installation extended with localized page aliases. A single page-id or concept would have many title columns for each language. (The localized title columns can be propagated by the existing database of inter-wiki language links). Furthermore since properties/relations have titles/page-ids; they could also be localized. Allowing you to query the shared structured dataset in your local language. Then something like external data extension will tie wikidata to all the current language wikis. This can be thought of as commons but for data. (likewise external to wikimedia wikis could use this structured data). This lets template authors concentrate on localized representation of the data (calling the native language properties) , articles authors can focus on the article (instead of huge seed of hard to maintain template data), and structured data folks can focus on importing data into the central shared repository. --michael Marcus Buck wrote: Lars Aronsson hett schreven: What is the best way to organize infobox templates for geographic places, the one used on the French, the Polish, or the Turkish Wikipedia? What are the most important features in use on other languages of Wikipedia, that my language is still missing? Are these questions of a kind that you sometimes ask yourself? If so, where do you go to find the answers? Are we all just copying ideas from the English Wikipedia? Or inventing our own wheels? Has anybody collected stories of how one project learned something useful from another one? As you are speaking of infoboxes and crosswiki, I want to chip in another thought: why do we actually place infobox templates on every single wiki? In 2007 I created some semiautomatic bot articles about municipalities on my home wiki. In 2008 they had elections and elected new mayors. So my articles mentioning the mayors were outdated. The articles in the main language of that country were updated relatively quickly, Mine are not yet. I plan to do, but who does that for all articles in all language editions? An example: Bavaria held communal elections in March 2008. Enough time to update infoboxes. The municipality Adelzhausen got Lorenz Braun as new mayor, replacing Thomas Goldstein. I checked all interwikis of the German article. Two had it right. Both were created after the elections. Four don't mention the
Re: [Wikitech-l] Interwiki recommendations
Michael Dale wrote:
We really need a wikidata type site.
A very easy and ugly workaround would be to store an image on
Wikimedia Commons, containting the letters Barack Obama and
having the filename President_of_USA.png. Next time change comes
to the White House, the image is replaced with a new version.
Then each language article could contain: The president of the
United States is [[file:President of USA.png]], and the right
name would automatically appear there. You'd need such images for
the names of all officicials, population of all cities, and so on.
(Don't scream. I copied this idea from the visitor counters of
mid-1990s websites, which were implemented by transcluding
images presenting the current number of visitors.)
From this ugly hack, it's easy to conceive that Wikimedia Commons
could distribute not only images but also text snippets and data.
An easy way to do this is to treat the template namespace
similar to the file namespace. Files (images) that aren't found
on the local wiki, are imported from Wikimedia Commons. If the
template {{president of the USA}} is not found locally, that
template would be sought on Wikimedia Commons.
With the right parameter setup and #switch: constructs, it could
be handled with just a few templates, e.g. {{president|USA}}
containing {{#switch: {{{1}}}|USA=Obama|Russia=Medvedev}}.
Of course, we'd need similar templates for Cyrillic and Arabic
scripts. But they could be named {{Президент}} and contain
{{#switch:{{{1}}}|США=Обама|Россия=Медведев}}, etc.
Of course, this kind of transclusion doesn't help you to write the
article about Barack Obama, where it says that he was elected
president of the United States in 2008 and took office in January
2009. But that is not a piece of text that needs to be
automatically updated.
It's interesting what would happen if the template imported from
Commons calls other templates, that do exist on the local wiki.
In programming language terms, it would imply that template
expansion has dynamic scope, just like Emacs Lisp, rather than
static or lexical scope (like most languages),
http://en.wikipedia.org/wiki/Scope_(programming)
--
Lars Aronsson (l...@aronsson.se)
Aronsson Datateknik - http://aronsson.se
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Re: [Wikitech-l] Interwiki recommendations
Hoi, I really like Semantic MediaWiki. But horses for courses. SMW is great for content where articles exist within one MediaWiki installation. It is TRULY project based. OmegaWiki is data that is indeed a separate database. The reason for a separate database is exactly because you do not want to create unnecessary dependencies. This is where OmegaWiki and Semantic MediaWiki are essentially different. GerardM 2009/2/3 Roan Kattouw roan.katt...@home.nl Gerard Meijssen schreef: Hoi, Have a look at this: http://www.omegawiki.org/Expression:Nederland This is structured data It can be shown in multiple languages. It does allow for interwiki links... It already works with MediaWiki ... So the question is, why re-invent the wheel ? I think the wheel is called Semantic MediaWiki in this case :P Roan Kattouw (Catrope) ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Re: [Wikitech-l] Interwiki recommendations
Lars Aronsson hett schreven: What is the best way to organize infobox templates for geographic places, the one used on the French, the Polish, or the Turkish Wikipedia? What are the most important features in use on other languages of Wikipedia, that my language is still missing? Are these questions of a kind that you sometimes ask yourself? If so, where do you go to find the answers? Are we all just copying ideas from the English Wikipedia? Or inventing our own wheels? Has anybody collected stories of how one project learned something useful from another one? As you are speaking of infoboxes and crosswiki, I want to chip in another thought: why do we actually place infobox templates on every single wiki? In 2007 I created some semiautomatic bot articles about municipalities on my home wiki. In 2008 they had elections and elected new mayors. So my articles mentioning the mayors were outdated. The articles in the main language of that country were updated relatively quickly, Mine are not yet. I plan to do, but who does that for all articles in all language editions? An example: Bavaria held communal elections in March 2008. Enough time to update infoboxes. The municipality Adelzhausen got Lorenz Braun as new mayor, replacing Thomas Goldstein. I checked all interwikis of the German article. Two had it right. Both were created after the elections. Four don't mention the mayor at all, and six still mentioned the old mayor. No wiki had bothered to update the information. It would be much easier, if we had a central repository for the data. We would place infoboxes in the central wiki. Each wiki then could fetch the data from the central wiki just as images are fetched from Commons and render the data into a localised infobox. That would be much more accurate than maintaining redundant info on potentially hundreds of wikis. Marcus Buck PS: And that would be interesting in regard to botopedias too. Volapük Wikipedia was massively critized for creating masses of bot content. With a central wiki for data creating articles for example for all the ~37,000 municipalities of France would essentially be reduced to creating a template that renders the central content into an article. Little Wikipedias could greatly benefit, if they just had to create some templates to make available info on hundreds of thousands of topics to the speakers of their language. It would be very basic, infobox-like information, but it would be information. ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Re: [Wikitech-l] Interwiki recommendations
Step 1 would be making interwiki transclusion not suck. Its been a long-standing back-burner project of mine. -Chad On Feb 2, 2009 8:05 PM, Marcus Buck w...@marcusbuck.org wrote: Lars Aronsson hett schreven: What is the best way to organize infobox templates for geographic places, the one used on the F... As you are speaking of infoboxes and crosswiki, I want to chip in another thought: why do we actually place infobox templates on every single wiki? In 2007 I created some semiautomatic bot articles about municipalities on my home wiki. In 2008 they had elections and elected new mayors. So my articles mentioning the mayors were outdated. The articles in the main language of that country were updated relatively quickly, Mine are not yet. I plan to do, but who does that for all articles in all language editions? An example: Bavaria held communal elections in March 2008. Enough time to update infoboxes. The municipality Adelzhausen got Lorenz Braun as new mayor, replacing Thomas Goldstein. I checked all interwikis of the German article. Two had it right. Both were created after the elections. Four don't mention the mayor at all, and six still mentioned the old mayor. No wiki had bothered to update the information. It would be much easier, if we had a central repository for the data. We would place infoboxes in the central wiki. Each wiki then could fetch the data from the central wiki just as images are fetched from Commons and render the data into a localised infobox. That would be much more accurate than maintaining redundant info on potentially hundreds of wikis. Marcus Buck PS: And that would be interesting in regard to botopedias too. Volapük Wikipedia was massively critized for creating masses of bot content. With a central wiki for data creating articles for example for all the ~37,000 municipalities of France would essentially be reduced to creating a template that renders the central content into an article. Little Wikipedias could greatly benefit, if they just had to create some templates to make available info on hundreds of thousands of topics to the speakers of their language. It would be very basic, infobox-like information, but it would be information. ___ Wikitech-l mailing list wikitec...@lists.wikimedia ___ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
