Wikimedia Meta-Wiki

Interwiki synchronization

This is an archived version of this page, as edited by 77.190.85.110 (talk) at 22:37, 17 November 2008. It may differ significantly from the current version.

This page is intended to host discussions between representatives from individual wikis regarding interlanguage links between Wikipedia articles.

Interlanguage links can often become entangled in conflicts. For example, if the page en:Emergency medical technician links to fr:Ambulancier, but fr:Ambulancier links to en:Paramedic, this is a conflict. This confuses the readers of Wikipedia and usually cannot be automatically maintained by interwiki bots (such as Interwiki.py). This inability to maintain links automatically frequently leads to lack of updates, stale links etc.

The task of synchronizing these links manually is hard, but challenging and interesting. A central hub for discussing this will foster further cooperation and concord between different language Wikipedias.

Encyclopedic truth is universal, so parallel ontology between Wikipedias in different languages is usually desirable, but it is important to emphasize that no Wikipedia can force its ontology on other Wikipedias, and any significant changes of this kind must be decided by consensus.

Future directions

In the future the technology of interlanguage links may significantly change thanks to the Interlanguage Extension; see A newer look at the interlanguage link. When this happens, the discussions about interwiki synchronization may move to the wiki that will be set up for this extension.

Automated analysis

The largest connected component in the graph of interlanguage links between articles contains well over 70'000 nodes (including over 3'000 from the English edition) and over 3'000'000 links. There are 24 more connected components with over 1'000 nodes. In total, there are about 60'000 inconsistent connected components in the graph. These figures are based on an analysis of snapshots from late August 2008.

It is next to impossible for a human to untangle the largest graphs anymore. Also, it is infeasible to process all 60'000 or so components, even if most of them are quite small. For that reason, an automatic meaning detection approach, even very imperfect, might be quite useful. An example of such automated analysis can be found here (a middle-sized component with approx. 300 articles and about 25 meanings), here are some more. The results show the identified meanings, the key nodes and links leading to semantic drift, and a complete set of links to remove to guarantee consistency. A description of the graph of meanings in the DOT language used by Graphviz is also provided. A batch edition based on the above results is possible.

Personally, I am strongly in favor of such an automated correction. Note that this would be a one-time action, which apparently has never been taken before. The 70'000+ component must have been growing for years: it contained about 48'000 articles in March 2008.

I'll gladly answer any questions regarding the idea, and upload more examples of analyzed components. If the suggested course of actions is approved by the community, I'd be glad to provide generated edit recommendations for all inconsistent components in a suitable format. I don't have the bot permissions nor the necessary experience, so I'm looking forward to a cooperation with a bot owner. I've performed the analysis for the graph of category interlanguage links too, and it requires similar action too.

This is an application of methods which I have developed during my PhD research, I'm currently in the process of writing two articles documenting the topology of the graph and the methods used to identify meanings. In short, the approach used here arranges the nodes in space using the force-based graph layout algorithm with a custom potential. Then, during the reconstruction of meanings, shorter links are considered to be more trustworthy.

Thanks, Bolo1729 22:41, 29 October 2008 (UTC)[reply]

Opening a new case

  • After creating the case page, add its name to the top of the list in the "Current discussions section".

Current discussions

Dollar

In Chinese (I don't know if there is another language like this), 銀圓 Just discuss the coin minted from the silver, no include the paper form or other forms.

Dollar, the currency of many countries

Discuss a name of the official currency in several countries.

Disambig

Some of the pages are defined as disambiguation.

Silver coin

Just discuss the coin minted from the silver, not including the paper form or other forms.

These are unclear

Currency

Discussion

Chinese

I think the correct interwiki of "Dollar" would be "元 (貨幣)" in Chinese. And the Chinese yuan, which is "元 (貨幣)" interwiki to, must interwiki to a non-exsit article of "元 (人民币)"。--Dingar 01:41, 15 October 2008 (UTC)[reply]

Is zh:元 (貨幣) really about currencies named "Dollar"? I don't know Chinese, but it seems that it is a general article about currency. It mentions Yen, Euro and others and Babelfish translates its title as "Yuan (currency)". --Amir E. Aharoni 17:50, 15 October 2008 (UTC)[reply]
Thank you! you are right. zh:元 (貨幣) include Yen, Euro, Dollar and others.--124.72.221.76 13:04, 20 October 2008 (UTC)[reply]
Sorry, I am forget to login yesterday. 124.72.221.76 is me.--Dingar 00:48, 22 October 2008 (UTC)[reply]
Thanks, but it is still unclear where these Chinese articles should link. en:Currency? --Amir E. Aharoni 14:02, 22 October 2008 (UTC)[reply]

ar:, fa: and ur:

The ar:, fa: and ur: all belong to the Dollar part. I am moving them there. --Amir E. Aharoni 13:18, 15 October 2008 (UTC)[reply]

Conclusion

Epistemology

  Done

In English and many other languages, "epistemology" (or the corresponding etymologically related word) means "theory of knowledge". In French, Spanish and Italian however, it has a different meaning, and is considered a branch of the philosophy of science (see w:fr:Épistémologie as opposed to w:fr:Théorie de la connaissance). This has caused problems about the interwiki links. I have tried to fix them manually on single Wikipedias, but bots keep updating the wrong links...

Looking at all the links, I could identify three main kinds of articles: Those about theory of knowledge, those about "Épistémologie" in the French sense, and disambiguating pages between the two meanings:

"Theory of knowledge"

"Épistémologie"

Disambiguation

  • just as help: in bosnian (the same in croatian) episteology has two differen meanings. For first as gnoseology (= theory of knowlege) and second as epistemology (= theory of science). The base for this two meanings is the different in sight of other languages. In our article (bosnian) we don´t explain the theory but the meanings of those two words. In few weeks i gonna translate the german article and to classify it. If there is need for translation of some croatian sentences (talk) let me know and i gonna translate it. Please use my bs.talk page. Best regards --Seha 20:19, 22 November 2008 (UTC)[reply]

Problematic case

Croatian has the following two articles:

Both seem to be on theory of knowledge. On the two corresponding talk pages this has been identified as a problem (on the talk page of the first there is a very recent discussion about that). Unfortunately, I can't understand enough Croatian to see where the discussion is going.

Klarigo: Temas, konjekteble, pri la sama distingo, kiu estis firme establita por la du fakoj en la rusa lingvo:
* эпистемология [epistemológija] okupiĝas pri studado de scioj, ilia strukturo kaj funkciado;
* гносеология [gnoseológija] esploras la procezojn/meĥanismojn de scio-akirado.
Tre ofte tiuj terminoj aperas adjektive: epistema (ruse: эпистемический = rilata al scioj) kaj gnoseologia (ruse: гносеологический = rilata al akirado/akumulado de scioj).
Tio estas tre baza karakterizo de la distingo. Pli reliefan komprenon oni povas ĉerpi el la artikolo "Эпистемология" en [ru.wikipedia.org Ruslingva Vikipedio].
Amike, -Filozofo (talk) 21:02, 12 August 2025 (UTC)[reply]

Discussion

I propose to relink all the articles as in the list above. Your comments are welcome. Marcos 14:31, 17 October 2008 (UTC)[reply]

i agree. Ca$e 15:06, 17 October 2008 (UTC) however, at least, in german there is in at least 99 of 100 cases not ambiguity of the phrasing "epistemologie". it almost always means epistemology. it is very very uncommon to mean épistémologie. in fact, it would in almost all contexts count as an error if a student wrote epistemologie and meant épistémologie. furthermore, both words are pronounced differently. i therefore see no need for the proposed disambiguation under epistemologie. the german word epistemologie simply means epistemology. hence, it was unaninmously agreed to redirect from epistemologie to epistemology when this was discussed. Ca$e 20:12, 20 January 2009 (UTC)[reply]

I noticed the problem with the Hebrew article some time ago and couldn't solve it myself, because i don't understand the subject. I cannot be sure about all the languages, but from what i can understand:

  • these are certainly or almost certainly correct: an, ar, bg, ca, cs, de, en, eo, es, fr, he, it, la, lt, pl, pt, ru, simple, sk, sl, sq, uk
  • i am not sure about tr - it mentions "bilim", which is "science", so maybe it is related to the French epistemology. But i might be wrong.
  • gl and oc should probably go to the Theory of knowledge part. I studied both languages a little and it seems that the description fits Theory of knowledge better.
  • bs:Epistemologija and sl:Epistemologija are not technically disambigs - they don't have the disambig template, which may cause minor problems to the interwiki bots. I left messages about it at the talk pages.

I also left messages in the talk pages of the Croatian articles. I can read Croatian a little, and they are indeed considering a merge or creating a disambig (razdvojba); maybe this will help them decide. --Amir E. Aharoni 22:21, 19 October 2008 (UTC)[reply]

I originally listed the Turkish article under "Theory of knowledge", since it mentions "Edmund Gettier", who is important for the theory of knowledge, but has no relation (as far as I know) to Épistémologie.
After reading Amir E. Aharoni's doubt about the Turkish article, I contacted my brother, who knows Turkish farely well. As he doesn't know much about philosophy, he couldn't say precisely which of the two issues the Turkish article is about. But he said that according to the article, "epistemoloji" studies the origin, nature and roots of knowledge. This sounds to me more like "theory of knowledge". However, the article also says: "Foucault uses this term to study the forms of knowledge and power structure". I don't know much about Foucault, but if he used a term etymologically related to "Epistemoloji", then probably it was "Épistémologie" and had the meaning it normally has in French. Additionally, the Turkish article says: "The term is also used in the sense of 'theory of knowledge' (bilgikuramı)." The use of "also" in that sentence (if my brother correctly translated it) indicates that the intended meaning in the rest of that article is not "theory of knowledge"...
In order to find out more about the issue, I left a message on the talk page of the Turkish article.
At any rate, it might be the case that the Turkish article (or other articles in other languages) treat both "Épistémologie" and "theory of knowledge" in one article (there is enough of a philosophical link between the two to make this sensible). In that case, I would favour linking the article with those articles with those articles, which disambiguate between the two concepts.
(Amir E. Aharoni says that the absence of disambiguation templates might confuse bots. Is that really the case? Don't the interwiki bots just look at interwikis? It would surprise me if they looked at templates too.) Marcos 17:02, 25 October 2008 (UTC)[reply]
They do look at templates and yield a warning to the bot operator. I am not a bot operator myself, but it may confuse someone who is. It is not a very big issue, but i do like cleanliness. --Amir E. Aharoni 13:07, 27 October 2008 (UTC)[reply]

Please check the automated analysis of this component. --Bolo1729 09:07, 30 October 2008 (UTC)[reply]

The automated analysis is no help, as it uses as its bases the wrongly interwiki links that have been reproduced in large quantity throughout different Wikipedias.
I have now changed all interwiki links according to my above proposals, with gl and oc changed according to Amirs proposal. I have linked from both Croatian articles to the theory of knowledge articles, but did not link back to either of the Croatian articles. Marcos 19:59, 22 November 2008 (UTC)[reply]

Update 2023

Note that the previous discussion is obsolete. Interwiki links are now centralized at Wikidata. On 3 May 2023, a decision was made at d:Talk:Q9471 to separate the related terms into three lexical cognates at d:Q9471 + d:Q2560959 + d:Q116930361, and all three are connected by d:Property:P460 ("this item is said to be the same as that item, but it's uncertain or disputed"). There is also a related term at d:Q30748859. The hope is that dividing the terms lexically instead of semantically, we can resolve recent conflicts related to the unstable or disputed semantics. Biogeographist (talk) 17:23, 5 May 2023 (UTC)[reply]


Brand and trademark

The interwiki linking of articles about trademarks and brands is pretty hard. They describe two main different, but somewhat overlapping concepts - "brand" as corporate identity in discussion of marketing and "trademark" as a legal term, which is separate, but quite related.

Here's how i tried to group the existing articles in different languages:

Brand

"Name, term, design, symbol, or any other feature that identifies one seller's good or service as distinct from those of other sellers."

Trademark

"Distinctive sign or indicator, used by an individual, business organization, or other legal entity, to identify that the products or services with which the trademark appears originate from a unique source, and to distinguish its products or services from those of other entities. Often indicated with ™, ℠ or ® (depending on legal status)."

Hard to understand

Wrong

This is definitely wrong, as this is just the disambiguation for the word "Serbian", but it appears on some articles.

Discussion

Can anyone please help with moving the links from the "Hard to understand" section to their right section? Of course, checking that the current grouping is correct would also be very helpful. --Amir E. Aharoni 14:00, 22 October 2008 (UTC)[reply]

Please check the automated analysis of this component. --Bolo1729 10:22, 30 October 2008 (UTC)[reply]

pl:Marka producenta

In fact this article describes both Brand and Trademark as if there one thing. The first line, translated to English, goes: "Brand, also known as manufacturer mark, trademark". I suggest removing them from interwiki linking until it'll get cut in two.


nl:Merk

As a native Dutch speaker, I understand your choice for nl:Merk's classification as "hard to understand". The article discusses various aspects of dealing with brands, including marketing and legal, but not very explicitly. One could split up the article, but that could destroy the "whole picture" of how brands fit into the (capitalistic) economy and how both aspects are inseparable: without identity it cannot be protected and without protection it fails as an identity. In the English Wikipedia, it could be argued that there should be a 'glue' article "Brand", a marketing article "Branding" and a legal article "Trademark". --nl:Gebruiker:Bdijkstra as 82.169.238.67 20:10, 22 October 2008 (UTC)[reply]

sq:Marka tregtare/sq:Marka tregëtare

Yes, sq:Marka tregtare and sq:Marka tregëtare are variant spellings of the same phrase, and I believe the former is correct (based on both a dictionary and google counts). They're talking about symbols and words that are intellectual property, not broader concepts like brand, so I suppose they correspond to "trademark". Unfortunately my Albanian isn't good enough to merge them myself, though I can stick a merge stamp on them. (Also, Emërtimet Tregtare translates literally as "trade names".) --sq:Përdoruesi:Steorra as 68.252.47.196 05:03, 24 October 2008 (UTC)[reply]

fr:Marque commerciale

fr:Marque commerciale clearly is about en:Trademark and its legal implications. Based on the title and most of the content, fr:Image de marque (literally "brand image") seems to be about en:Brand, although it links to en:Brand management. --Latebird 22:42, 1 November 2008 (UTC)[reply]

Conclusion

Remembrance day

Starting from ksh:Volkstrauertag via de:Volkstrauertag we come to:

  1. en:Volkstrauertag
  2. en:Remembrance Day
  3. en:Armistice Day

These days are all on the same date but are having different meanings. Luckily, en:Halloween was not number 4.

Solution: Isolate the meanings.

Problem 1: Maybe some other languages have still other meanings in the cluster which I am currently not aware of.

Problem 2: I can only handle an incomplete set of languages/scripts, so I cannot do it alone.

Discussion

  • HE: (checked, and didn't find any articles)
  • Volkstrauertag is on a different date.Andreas 15:49, 19 November 2008 (UTC)

(check and report more languages)

Conclusion

elementary-primary-comprehensive-school-education-etc

There are several starting point in various wikipedias, all having to do with elementary/basic/secondary/comprehensive/collaborative schools/schooling/learning/education.

They connect:

  1. en:Comprehensive school
  2. en:Elementary school
  3. en:Gastropoda — (see #Group 2)
  4. en:Gymnasium (school)
  5. en:High school
  6. en:Liceum ogólnokształcące — (see also #Group 1.A)
  7. en:Lyceum — (see #Group 1)
  8. en:Primary education
  9. en:Primary school
  10. en:Secondary education
  11. en:Secondary education in France#Lycée
  12. en:Slug — (see also #Group 2.B)

One of the problems apparently is that, school systems in different countries just do not compare. Likely, most school types apply to a much smaller set of countries (and thus likely, languages, too) than current interwiki links try to make us believe. --Purodha Blissenbach 21:59, 2 November 2008 (UTC)[reply]

Group 1

Starting from en:Lyceum gives a clean group:

  1. ca:Liceu (antiga Grècia)
  2. cs:Lyceum
  3. de:Lykeion
  4. el:Λύκειο
  5. en:Lyceum
  6. eo:Liceo
  7. es:Liceo
  8. fr:Lycée
  9. he:ליקיאון - Aristotle's school near Athens, est. 335 BCE
  10. hu:Líceum
  11. it:Liceo di Aristotele
  12. ja:リュケイオン
  13. nl:Lyceum
  14. pl:Lykeion
  15. pt:Liceu
  16. ru:Лицей
  17. sk:Lykeion
  18. sr:Лицеј
  19. sv:Lykeion
  20. tr:Lykeion

thus cutting all links from the others to this group might be a helpful step. --Purodha Blissenbach 23:50, 2 November 2008 (UTC)[reply]

Group 1.A

Starting from it:Liceo, things already get messy:

  1. cs:Lyceum
  2. de:Lyzeum ← not good: disambiguation. Links ignored.
  3. el:Λύκειο
  4. en:Lyceum
  5. fr:Lycée
  6. he:ליקיאון - Aristotle's school near Athens, est. 335 BCE
  7. hu:Lyceum → hu:Líceum
  8. it:Liceo
  9. ja:リセ
  10. pl:Liceum ogólnokształcące
  11. sr:Лицеј
  12. sv:Lykeion
  13. ca:Liceu (antiga Grècia)
  14. eo:Liceo
  15. es:Liceo
  16. nl:Lyceum
  17. pt:Liceu
  18. ru:Лицей
  19. sk:Lykeion
  20. tr:Lykeion
  21. uk:Середня загальноосвітня школа

Group 2

Starting from any of the following yields no problem links either:

  1. ar:بطنيات القدم
  2. bg:Коремоноги
  3. bs:Puževi
  4. ca:Gasteròpode
  5. cs:Plži
  6. da:Snegl
  7. de:Schnecken
  8. en:Gastropoda
  9. eo:Gastropodo
  10. es:Gastropoda
  11. et:Teod
  12. fa:شکم‌پایان
  13. fi:Kotilot
  14. fr:Gastropoda
  15. he:חלזונות - Gastropoda
  16. hr:Puževi
  17. hu:Csigák
  18. io:Gasteropodo
  19. is:Sniglar
  20. it:Gastropoda
  21. ja:腹足綱
  22. ka:მუცელფეხიანები
  23. ko:복족류
  24. la:Gastropoda
  25. lt:Pilvakojai
  26. lv:Gliemeži
  27. mk:Полжав
  28. nds-nl:Slekken
  29. nl:Slakken
  30. nn:Sniglar
  31. no:Snegler
  32. pl:Ślimaki
  33. pt:Gastrópodes
  34. qu:Wiksachaki
  35. ru:Брюхоногие
  36. simple:Gastropoda
  37. sk:Ulitníky
  38. sl:Polži
  39. sq:Kërmilli
  40. sr:Пужеви
  41. sv:Snäckor
  42. tr:Karından bacaklılar
  43. uk:Черевоногі
  44. zh:腹足纲

--Purodha Blissenbach 00:26, 3 November 2008 (UTC)[reply]

Group 2.B

starting from en:Slug yields a duplicate each for almost all languages, and collects all members of the above #Group 2, too. --Purodha Blissenbach 00:39, 3 November 2008 (UTC)[reply]

Discussion

I've uploaded an automatic analysis for this area, it's here. It is based on database snapshots from late August 2008.

At least three "bottlenecks" in the graph of meanings can be identified:

removed from en:Secondary education in France and fr:Lycée en France, changed interwiki to fa:لیسه (حلزون) elswhere. Andreas 15:42, 19 November 2008 (UTC)

Meaning 12

de:Klammeraffe ("@-sign") should be isolated. It is unrelated to en:slug (disambiguation). The link is caused by homonym hu:Csiga which means both "@" and "slug" as it appears to me. (Imho linking disambiguations other than literals (abbreviations, figures, codes) works usually only between related languages having very compatible sets of homonyms) --Purodha Blissenbach 21:43, 3 November 2008 (UTC)[reply]

I removed the interwikis from these pages. Andreas 14:23, 19 November 2008 (UTC)

Conclusion

Google

Currently, we have the situation where the search engine Google and the company Google Inc. are totally mixed up in the various interwikis. These need to be cleaned up and separated.

Search engine

Company

Unsorted

Wrong

Discussion

Here's an automated analysis of this component. Indeed: there are lots of incorrent interlanguage links between the meanings containing en:Google (meaning #9) and en:Google search (meaning #1). Interwikis to/from the two Bengali articles: bn:গুগল (সংখ্যা) and bn:গুগল (কোম্পানি) seem to be one of the core reasons of inconsistencies. --Bolo1729 11:25, 17 November 2008 (UTC)[reply]


Some languages don't distinguish between different google activities: they have one article for the google company, the search engine, and all other activities (gmail, gdocs, etc'). I suggest to classify such articles under a new category or under the google-company. What do you think? TaBaZzz 18:47, 6 December 2008 (UTC)[reply]

Going through the unknowns, I get to the following:

(still working on it as you see) - Andre Engels 08:44, 9 January 2009 (UTC)[reply]

Soldier - Military personnel

These two categories are interlinked:

In German, they do not seem to make a destintion between them, or they only have:

which is "person in the military" but actually, the category seems not to contain civil personnel, and only persons.

--77.190.85.110 22:36, 17 November 2008 (UTC)[reply]

I've run an analysis for this case. The graph of meanings clearly shows that en:Category:Military personnel is the central category in this case. There is quite a lot of confusion between that and en:Category:Generals, as well as en:Category:Soldiers. --Bolo1729 10:16, 18 November 2008 (UTC)[reply]
Civilian military persons? Hmmm... Actually, de:Kategorie:Militärperson includes the following subcategories containing (at least partly) civilians and/or noncombattants:
According to the respective articles, a en:Solder is a combattant with land forces only, while a de:Soldat is is a combattant in any military force. Because of that difference in scope, matching categories are not to be expected on that level. --Latebird 22:49, 5 December 2008 (UTC)[reply]

Archived discussions

September 2008

October 2008

See also

Retrieved from "https://meta.wikimedia.org/w/index.php?title=Interwiki_synchronization&oldid=1282529"