Grants talk:IdeaLab/Reform of citation structure for all Wikimedia projects

Who wants to manage this?

I just listed a lot of projects each of which could take years of development. I cannot manage even one of these nor could I manage the central coordination of all of them together. Anyone who wishes to address any of this can join individual projects or oversee all of them. I would love for someone to turn this into a grant proposal for themselves.

I would like to give community input into all these ideas, because as a Wikimedia community member I have problems with the citation structure and I want it to be developed and improved.

Anyone who wishes to talk to me can do so here or email me for a video conversation on Skype or wherever. Blue Rasberry (talk) 14:10, 18 March 2014 (UTC)

Individual aspects

Each of the pieces can be done independently. I am supportive of most of them. Getting them all done at once will be hard. Doc James (talk · contribs · email) 23:54, 18 March 2014 (UTC)

Everything to do with Citevar

"The creators of these tools intended to generate consistent citations" -- Really? That would seem to directly to contradict Citevar. There are variation in citation styles and the differences the styles that different tools generate is completely consistent with Citevar. Since there is no citation "house style", what is needed is flexibility in the output of the different tools so that each matches the others format, not that each tool should produce exactly the same format. Boghog (talk) 22:48, 21 March 2014 (UTC)

I would recommend using CSL/citeproc to accomplish this. Then the problem would basically be reduced to converting the wikidata item to the citeproc-json format. Klortho (talk) 07:53, 22 March 2014 (UTC)

Boghog, what you are saying is what I intended to express. I want to encourage intentional variation but stop unintentional variation. I do not know what Klortho is saying but I think he is talking about structuring data in a consistent way in the database so that everyone draws from the same datapool and would by default have consistent tool output unless someone choose to vary from the default. Blue Rasberry (talk) 15:48, 5 May 2014 (UTC)

That's right. I'm suggesting separating the content from the presentation -- a very common pattern in, for instance, web development these days. The "citeproc-json" is a normalized, consistent "hub" data format that could be fairly easily generated from wikidata. Then, CSL (Citation Style Language) could be used to convert this data into formatted citations according to the "house style", however that would be defined. CSL is used more and more widely these days in the publishing industry. For example, most of the bibliography managers like Mendeley and Zotero use it to generate formatted citations for authors to insert into their manuscripts. There are thousands of different styles already defined, and they are all maintained in the open, on GitHub. Klortho (talk) 04:21, 6 May 2014 (UTC)

I follow the first two sentences and agree with that those. I get a bit lost after that but I expect I agree with all the rest and that I want things to be that way. Blue Rasberry (talk) 20:44, 6 May 2014 (UTC)

┌────────────────┘
I think there are layers to this. It seems everyone agrees that the following three hypothetical articles are following policy and each is using a different citation policy:

1. Article uses no citation templates of any kind, but the citations all follow a similar format, which may be a different format from any other article.
2. Article uses the Citation template.
3. Article uses citation templates in the en:Help:Citation Style 1 (CS1) family.

But in the past the documentation for all the varieties of citation template emphasized stating the definition of the parameters, and didn't contain much information about how to write the value of the parameter. Such information would be provided in printed style guides, but was lacking in the template documentation. For example, should an institutional author listed with one of the types of author parameter, or should an institutional author always be treated as a publisher? Should an author's name be given in full, or should the first and middle name be reduced to initials? Recently en:Help:Citation Style 1 has been edited to begin to fill in some of these details, but at least en:User:Boghog resists this, indicating that templates in the CS1 family may be used to implement a variety of styles.

Another point is that inserting any machinery within an article to retrieve a citation from some central repository, rather than just writing the citation within the article without any templates or similar machinery, would be viewed as changing the style of the article and would probably be resisted on the basis of CITEVAR. Jc3s5h (talk) 21:16, 28 May 2014 (UTC)

Ok, please forgive me if I ramble a bit, but really, I CBA about that. It strikes me that English Wikipedia's preoccupation with CITEVAR is not likely shared by the many other wikis and languages where editors are struggling with the more basic issues of getting verifiable content, getting supporting citations for unreferenced statements, etc. Even if a substantial part of that one wiki considers such stylistic issues more important than verifiability (which to my mind is an unsustainable position) it would still be the height of irresponsibility to allow that to impede the portability of citations between wikis.

There is great effort being expended elsewhere in building massive bibliographies. Until it can be fully automated, there is little prospect that we would improve on the reach or accuracy in the work of CrossRef, PubMed, HathiTrust, JSTOR, WorldCat, OpenLibrary, etc., even though each of these has substantial flaws and limitations. Rather, we should seek to ease the wiki-editors' use of the metadata these others have collected. Much metadata about source publications is numeric, and so usable in all languages. That part is the key to laying physical or virtual hands on the source and should be our priority. Another large portion is in the language of the specific publication, so that translation is informative but not essential. That portion should be easily dealt with using librarians' existing tools. The larger job of providing translation of descriptive parts (e.g. titles) into local languages will likely be done only as the citing articles are translated or even transliterated. A model somewhat like wiktionary for multilingual representation of these translations might be helpful, but such wiki-generated translations should be kept quite apart from the externally verifiable metadata. These will be far more likely to be subject to disputes and to vandalism than the basic data.

It is way past time that we segregated citations from the bibliographic data about the work cited. An inline citation is intended to say that the citing editor based a statement on a specific part of the cited work. It makes very little sense to have every citing article (on any wiki) replicate the bibliographic data with variations, when all they really need is a page number (or similar), a robust link to a trusted bibliography, and perhaps an editnumber to attribute the addition of that citation to a specific edit (this last a longstanding omission addressed only poorly by Wikiblame). That of course says nothing about presentation to the reader, just about capturing the linkage between an assertion in the wikitext and the part of a source which is cited in support. Frankly, if that linkage is solidly captured, I have little to no concern about the format of presentation to the reader. Leaving that up to article editors would then be only a matter of style. LeadSongDog (talk) 21:38, 12 June 2014 (UTC)

this proposal is unlikely to go anywhere with the current state of citevar and the emotional attachment some have to their favorite formats. go look at all the edit wars around citationbot and some editor's rabid preference for their vancouver varient.Chris Capoccia (talk) 19:33, 20 January 2015 (UTC)

Wikimedia hackathon

I think the Zürich Hackathon 2014 would be a good opportunity to move this forward, e.g. by detailing the specs for possible subprojects. -- Daniel Mietchen (talk) 04:07, 22 March 2014 (UTC)

References Namespace

Worth mentioning is a potential addition to the list of projects, or at least an example for context / history. In Wikipédia Français there is a "Référence" namespace for managing references. Some examples:

• fr:Référence:Dictionnaire_historique_de_la_langue_française
  • Usage list of pages
• fr:Référence:Dictionnaire_historique_des_rues_de_Paris
  • Usage list of pages
• More examples
  • List of all pages in References namespace

This practice is worth mentioning because it illustrates the experimentation and range of solution development in the pursuit of improving the way citations currently work, across domains. It would be great to elaborate on how the list of current projects and all previous or ongoing efforts have been an invaluable contribution to continued improvement. They make the case clear that with only a small amount of coordination, it is possible to encourage growth that is critical for the efficacy of the Wikimedia movement, but also consistent with various relevant histories and in a discernible narrative.

- Mattsenate (talk) 07:23, 25 March 2014 (UTC)

An interesting dvelopement with this solution would be to have a database of citations that can be processed and read by Wikipedia articles. (Sort of Wikiquote for references). Articles would not automatically include the full set, biut Wikidata could attach to a topic a list of references that can be queried to find accurate data with some filters by language, such as bibliographic details like the ISBN or local editor; then a template in Wikipedia would include a reference found in Wikidata for inclusion in the article by its unique ID, and a tool would allow searching about related references. Then instead of using complex and lengthy "<ref>" tags evereywhere in articles we could insert a single "{{GlobalRef|id}]" in the article, the template generating the actual ref data to render, possibly with the help of a local namespace like "Reference:" in French Wikipedia, plus specific tools to help manage the collection and filters (prefered languages, country of edition, maximum number of references, some qualifying data for evaluating the relevance such as a notation of sources, open access or no-pay access, date of publication for news and event reports, state of freshness or consolidation...) from Wikidata. verdy_p (talk) 00:36, 16 June 2014 (UTC)

"A free and universal bibliography for the world"

Include the WikiScholar proposal in background.

A fact is only as reliable as the ability to source that fact, and the ability to weigh carefully that source.^[1]

Mattsenate (talk) 05:15, 26 March 2014 (UTC)

This narrative flows well alongside the technical development from mw:Semantic Mediawiki to WikiData: introducing the possibility, and really, the inevitability of these changes.

Talk about this general subject at Wikimania 2014 in London

I have a proposal to discuss this at wm2014:Submissions/Reform of citation structure for all Wikimedia projects during Wikimania 2014 London. Blue Rasberry (talk) 16:01, 5 May 2014 (UTC)

Wikimania panel: Reform of citation structure for all Wikimedia projects, Video, Author/Speaker :

• Moderator: Lane Rasberry - Consumer Reports
• Max Klein - Wikidata Sources Taskforce
• David Cuenca - Wikisource User Group
• Adam Becker - PLoS (Open Access Publisher)
• Megan Wacha - Research and Instruction Librarian for the Performing Arts at Barnard College of Columbia University
• James Forrester - Visual Editor

See also d:Wikidata:Referencing improvements input. --Atlasowa (talk) 10:18, 28 November 2014 (UTC)

Umbrella page for these efforts

I created an umbrella page (WikiCite) including a history of this and related efforts.--DarTar (talk) 02:05, 6 September 2016 (UTC)