This thread seems really to be dealing with two distinct issues: coverage and quality.

On the coverage side, the best work I know of is from Brent Hecht and Darren Gergle. Especially [1] but also [2,3]. The English edition may be the greatest in size, but in reality most other languages contain many unique articles not found in English. All the editions to some extent suffer from a self-focus bias where regions where the language is spoken are better represented than regions where the language is not spoken.

Quality of existing articles is completely separate from coverage and is harder to evaluate at scale. I understand that some of the work at GroupLens (Minnesota) around SuggestBot is trying to develop metrics that can measure quality at scale in different languages (and working to develop/use different metrics in different languages as appropriate because not all editions have the same conventions for referencing styles, etc.). I don't have a specific reference here, but would say keep an eye on this area.

~Scott


[1] Hecht, B., & Gergle, D. (2010). The Tower of Babel meets Web 2.0: User-generated content and its applications in a multilingual context. Proceedings of the 28th International Conference on Human Factors in Computing Systems (pp. 291-300). New York, NY, USA: ACM. doi:http://doi.acm.org/10.1145/1753326.1753370

[2] Hecht, B., & Gergle, D. (2009). Measuring self-focus bias in community-maintained knowledge repositories. Proceedings of the Fourth International Conference on Communities and Technologies (pp. 11-20). New York, NY, USA: ACM. doi:http://doi.acm.org/10.1145/1556460.1556463

[3] Bao, P., Hecht, B., Carton, S., Quaderi, M., Horn, M., & Gergle, D. (2012). Omnipedia: Bridging the Wikipedia Language Gap. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 1075-1084). New York, NY, USA: ACM. doi:10.1145/2207676.2208553



On Tue, Jun 10, 2014 at 7:34 AM, Anders Wennersten <mail@anderswennersten.se> wrote:
My starting point have been the newly created articles on svwp. They will represent the usual bunch of football playser, tv-stars, computergames, films  etc where svwp are behind most versions but where enwp is excellent. The interesting comparisons comes from the next levels of articles that can be almost anything, a footballstadium in Kazan Russia, an albanian poet, a church in Venize, a specie with unclear taxonomy, the american solider who perhaps deserted etc. In these cases I only often find a corresponding article in enwp, but also very often (around 20%) I find it in another version and no presence in enwp.

And when enwp is not giving me support, I most often find support in eswp and frwp, sometimes in dewp, but almost never in ptwp. For exemple taxanomical threes  with name in native and latin is about the weakest in ptwp. But I can be wrong and I would love to be part in a more complete research on Q comparisons for the different versions


Anders





Juliana Bastos Marques skrev 2014-06-10 14:06:
This topic comes in handy for my research on Featured Articles in WP:PT. Maybe some of you may remember my request a little while ago about studies on Wikipedias other than English. Well, not that I believe that the Featured Article requirements are a good evaluation per se, in terms of quality of content.

Anders, what are the articles you evaluated? I'm curious to find out what was so bad in the Portuguese Wikipedia. Indeed, there are many problems there, but I'm surprised to hear that it looks so bad. I know it's a drop in the ocean, but I've been fixing some new articles that are translations from bad English ones - which look good, but analyzing the content reveals many problems.

Juliana.


On Tue, Jun 10, 2014 at 8:29 AM, Anders Wennersten <mail@anderswennersten.se> wrote:
Thanks for answer

Your answer confirm my "fear", that focus is almost completly to en:wp and how it is compared with an ideal perfect Q

My interest and what I believe the movement need before we dig into next round of strategy round is
*what versions are dysfunctional. These represent a risk for the movement as they can jeoprdaize the brand name, as they are not living up to basic Q (and NPOV)
*what can we learn from each other, why are some better in some aspects and worse in others?

I would recommend a research approach much more basic just collecting some few data on each version (and forget about enwp)

Anders



Heather Ford skrev 2014-06-10 13:09:
Hi Anders,

Yes, it's a great question! Mark Graham and I are currently working on a project around how to determine quality within and between Wikipedias and I've been looking around for literature. I'm only just starting the literature review but I've found some interesting studies by Callahan & Herring (2011) [1] and Stvilia, Al-Faraj, and Yi (2009) [2]. The majority of quality studies, we find, have been done on English Wikipedia (starting with the famous 2005 Nature study) but there have been few studies that assess of quality between languages. If you find anything else, let us know!

Thanks!

Best,



On 10 June 2014 07:58, Anders Wennersten <mail@anderswennersten.se> wrote:
(reposted from Wikimedia-i)

I have several times asked for a professional quality study of our different language versions, but not seen it exist or being done, perhaps you know more on this list?. before we start the strategy work I  believe we should have basic facts on the table like this one

I therefor list here my subjective impression after daily looking into the different version for 5-15 articles (new ones being created on sv.wp) (I list them in order how often I use them to calibrate the svwp articles).

enwp- a magnitude better then any other. main weakeness are articles on marginal subjects that seems to be allowed to exist there, even if rather bad, and without templates (noone cares to patrol these?)

eswp - a very  good version, which in the general discussion are not getting appropriate credit

dewp - good when the articles exist, but many serious holes. Is the elitist way of running it, discouraging new editors in non obvious subjects (that after time passes gets very relevant)?
frwp - also good, but somewhat scattered quality both in coverage and the different articles (even in same subject area)
nlwp - very good coverage in the geographic subjects, decent quality on articles but limited "world" coverage in areas like biographies
itwp - good articles but a bit italiancentered,

nowp - small but decent articles. Their short focused articletext sometimes give more easyaccessed knowledge then an overly long one in other languages

ptwp - the real disappointment. it is among the top ten in volume and accesses but clearly missing a lot, and even existing articles are uneven. I now use it even less then Ukrainian and Russian which I use very seldom as the different alphabet makes it hard to understand the article content

dawp,fiwp and plwp -Ok but only used by me for articles related to the country

(arabic, chinese and japanese I almost never use, too complicated)

(I also use some smaller ones like sqwp , in these versions I have seen serious quality problems not to be found in any of the above ones, I am not sure they even have basic patrolling in place)

Anders

_______________________________________________
Wiki-research-l mailing list
Wiki-research-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l



_______________________________________________
Wiki-research-l mailing list
Wiki-research-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l


_______________________________________________
Wiki-research-l mailing list
Wiki-research-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l




--
www.domusaurea.org


_______________________________________________
Wiki-research-l mailing list
Wiki-research-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l


_______________________________________________
Wiki-research-l mailing list
Wiki-research-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l




--
Scott Hale
Oxford Internet Institute
University of Oxford
http://www.scotthale.net/
scott.hale@oii.ox.ac.uk