Help talk:Citation Style 1/Archive 15

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

There is a specification for the |bibcode= parameter here. I think it would be worth checking for invalid Bibcodes.

This request was motivated by a discussion that started at User Talk:Citation bot. Pinging Lithopsian. – Jonesey95 (talk) 16:56, 28 April 2016 (UTC)

The bibcode isn't super stringent in its assignment, e.g

books

} 17:03, 28 April 2016 (UTC)

That sounds like a good place to start. – Jonesey95 (talk) 17:30, 28 April 2016 (UTC)

In the sandbox using the arXiv bibcode above:

Pass: year and length ok.

Bibcode:2015arXiv151206696F

.

fail: length too short.

Bibcode:2015arXiv151206696. {{cite book}}: Check |bibcode= length (help

)

fail: length too long.

Bibcode:2015arXiv151206696FF. {{cite book}}: Check |bibcode= length (help

)

fail: first 4 not a year.

Bibcode:2o15arXiv151206696F. {{cite book}}: Check |bibcode= value (help

)

fail: other than letters, digits, and dots.

Bibcode:2015arXiv...1.!.l.. {{cite book}}: Check |bibcode= value (help

)

—Trappist the monk (talk) 22:29, 28 April 2016 (UTC)

According to this reference it is possible to see colons and percent signs as well as letters, digits, and dots in a bibcode. (Actually, it doesn't say what characters can appear, but it describes circumstances in which those two additional characters will appear). —David Eppstein (talk) 22:51, 28 April 2016 (UTC)

For the last check, & is an accepted character. I've never seen : or % however, but I suppose they might happen. Or are old characters. Because no authors usually is a . in the last character (see

books

} 00:27, 29 April 2016 (UTC)

Do you mean by this that you have seen (and so can provide an example) of a case where the character in the 19th position is an ampersand?

Code tweaked to accept dot in 19th.

—Trappist the monk (talk) 00:40, 29 April 2016 (UTC)

Pass: Dot as 19th character.

Bibcode:2003xmm..pres...36.

—Trappist the monk (talk) 00:42, 29 April 2016 (UTC)

(edit conflict) though I see that edit conflict detection is still broken...

That's a bit of a bother. Looks to me like the ampersand is allowed in the journal abbreviation somewhere in the five characters following the year (character positions 5–9) A&A = Astronomy & Astrophysics. Both documents agree on this. At character position 19, one document allows letters, colon, and percent; the other, just letters. But then there is this, 14 characters, no year, disallowed character, yet the link works:

fail: doi.

Bibcode:10.1086/378243. {{cite book}}: Check |bibcode= length (help

)

but not all dois:

fail: doi.

Bibcode:10.1051/0004-6361:20078467. {{cite book}}: Check |bibcode= length (help

)

What is the point of having a standard if the standard is ignored? Yeah, that's a rhetorical question ...

I've tweaked the code so that character position 5 only accepts letters (abbreviations are left justified) 6–9 accept letters, ampersand, and dot; character position 19 will accept letters and colon. I chose to disallow percent because '%' is the first character of a url encoding. We have not been url encoding bibcodes so I would expect that were there bibcodes in the wild that use the '%' in the 19th position, we would have heard about that by now. If that turns out to be an issue, I think that we can turn on url encoding without too much trouble.

Pass: year, journal abbreviation, and length ok.

Bibcode:1994A&A...282..663S

.

Pass: Colon as 19th character.

Bibcode:1994A&A...282..663:. {{cite book}}: Check |bibcode= value (help)CS1 maint: extra punctuation (link

) – 19th character replaced with a colon as a detector test; the link doesn't work

—Trappist the monk (talk) 00:37, 29 April 2016 (UTC)

Passing the doi in the bibcode is more or less API abuse. It's not a bibcode, but sometimes it will work because

books

} 01:05, 29 April 2016 (UTC)

Also, colon as the last character, I've never seen that, and you're basing yourself on the refcode (bibcode's predecessor) documentation from 1995.

books

} 01:10, 29 April 2016 (UTC)

I expect that the first character can be limited to the numbers 1 or 2, since the four-digit year of publication of a journal article should always start with 1 or 2. That's just me telling a story, though. I'll scrub through some articles to see if I can find Bibcodes that would not meet these criteria but are listed in the database as valid. – Jonesey95 (talk) 01:54, 29 April 2016 (UTC)

We might be able to limit the first two characters to "17", "18", "19", and "20". Were there journals in the 1600s?

Some valid edge cases I found in articles:

Bibcode:1981NVS...C......0K

(lots of periods, just for fun).

I looked through a few hundred Bibcodes in a couple dozen articles and saw only letters and ampersands and periods in the journal name portion (although it is possible for numbers to be present, according to the conference list), and only letters and numbers and periods in the remainder of the identifier. I did not see any colons or percentage signs, FWIW, and those appear to be allowed only in REF_CODE, not in Bibcode. I think we should be OK with the check as it is written, but without the colon, and accepting only 17/18/19/20 as valid centuries (and maybe 16?) – Jonesey95 (talk) 02:38, 29 April 2016 (UTC)

The earliest thing indexed in ADSABS is from 1054 (Chinese reports of the

books

} 10:18, 29 April 2016 (UTC)

You did not answer my question about ampersand in the 19th position. Is an ampersand allowed in the 19th position?

I have constrained the allowed dates to the range of 1000 – next year so:

Fail: year too early.

Bibcode:0999A&A............ {{cite book}}: Check |bibcode= year (help

)

Pass: year ok.

Bibcode:1000A&A............

Pass: this year.

Bibcode:2016A&A............

Pass: this year+1.

Bibcode:2017A&A............

Fail: this year+2.

Bibcode:2018A&A............

—Trappist the monk (talk) 12:34, 29 April 2016 (UTC)

The ampersand is not allowed in the last position. AFAIK, it's only allowed in the journal code following the year. It will also never be the first or last of the journal code. So basically ampersand is allowed in positions 6-8.

books

} 23:01, 29 April 2016 (UTC)

I've removed the colon as a permitted 19th character and refined the journal abbreviation test so that it detects an ampersand-dot combination when the ampersand is in character position 6, 7, or 8 (not allowed elsewhere):

Pass: Ampersand not followed by a dot.

Bibcode:2016A&A............

Fail: Ampersand followed by a dot.

Bibcode:2016A&............. {{cite book}}: Check |bibcode= journal (help

)

Fail: colon as 19th character.

Bibcode:2016A&A...........:. {{cite book}}: Check |bibcode= value (help)CS1 maint: extra punctuation (link

)

Fail: Ampersand as non 6th-8th character.

Bibcode:2016A&A.........&.. {{cite book}}: Check |bibcode= value (help

)

—Trappist the monk (talk) 00:09, 30 April 2016 (UTC)

I think this is good to go now. Should probably populate its own category initially ('citations with bibcode errors' or some such), and then maybe a general category later once we cleaned up the first bunch.

books

} 12:55, 1 May 2016 (UTC)

The bibcode errors will be categorized into Category:CS1 errors: bibcode. We haven't switched specific error categories to a general category for other identifiers even though those specific identifier categories are now mostly empty, so I see no reason to do what with bibcode.

—Trappist the monk (talk) 16:02, 1 May 2016 (UTC)

Where we cite a work, or journal, which has a corresponding Wikipedia article, we can make the title a link; and we can use |authorlink= if there is an article about the author. What if there is no article, but there is a Wikidata item? How might we include, and display, the "Q" value, and link to the item? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:50, 13 April 2016 (UTC)

I suppose d: would work?

Author. Title. {{cite book}}: |author= has generic name (help)

EDIT: corrected the citation to comply with CS1.

72.43.99.130 (talk) 19:43, 13 April 2016 (UTC)

More particularly, linking to a "Q" value gives:

Author. Title. {{cite book}}: |author= has generic name (help)

However, that does not indicate to the user that the target is Wikidata, not Wikipedia. Consider also the output of {{

Red Wikidata link}}. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits

19:58, 13 April 2016 (UTC)

I assume that links are used in templates in order to aid verification. The link host is immaterial, as long as it provides a path to verification. A properly filled in Wikidata item will do that. I don't see a pressing need to identify the host. 65.88.88.76 (talk) 22:11, 13 April 2016 (UTC)

From the viewpoint of template design, we wouldn't want to use |authorlink= anymore but some new parameter specifically for wikidata ids. Otherwise we'd have one parameter that accepts data from two different domains (therefore the citation template — an input form — would fail the

principle of least surprise in user interface design. But even if we ignored these concerns, why would we even want to link the author's name in such a case? What makes the pros overcome the cons? I certainly see some value in such a link but I also think a convincing argument could be made that it would be a form of overlinking. Even if we did have a whole new parameter for Wikidata ids that displayed separately from the author name (like we do with |doi= and other document identifiers), a case for overlinking could still be made based on the idea that it's adding clutter. Jason Quinn (talk

) 07:11, 14 April 2016 (UTC)

I think that Jason's analysis is sound. Although it is clearly informative to be able to find other publications by the same author, that's not the purpose of a citation, which is simply to identify a source. I've long thought that authors, journals, etc. should not be wikilinked in citations: the only link should be to the source if it's online. Peter coxhead (talk) 09:34, 14 April 2016 (UTC)

I disagree (that is, I agree with the first half of Jason's analysis, about FNF & least surprise; but not that this would involve over-linking); but that still leaves the question of how to indicate the existence of a Wikidata item about an author or work. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:04, 14 April 2016 (UTC)

The citation templates are not database templates, neither are they input forms, although they do have a formalized design. They do exist to provide a standardized way for presenting sources that verify text in the article. Contributors can be agnostic about whether a link that aids verfication (for example, to a page that lists author works, and provides a path to them) is from Wikipedia or Wikidata: just use whatever link exists to do that. I doubt that readers (the vast majority of Wikipedia users) care about the provenance of the material that verifies the information. We need to have a link to information that verifies the citation. We don't need to parse every single step in the path to final verification. I don't include Wikisource links in this, since Wikisource is more akin to an online publisher. 72.43.99.146 (talk) 15:15, 14 April 2016 (UTC)

If there were a single link, as e.g. to a library catalogue item, which then gave ID links to all the authors (of which there can easily be 15+ in some scientific papers), plus the work (book, journal, or whatever), then I would support the proposed addition. Consider a citation like the following, which I recently used in an article: Miller, Jeremy A.; Carmichael, Anthea; Ramírez, Martín J.; Spagna, Joseph C.; Haddad, Charles R.; Řezáč, Milan; Johannesen, Jes; Král, Jiří; Wang, Xin-Ping; Griswold, Charles E. (2010), "Phylogeny of entelegyne spiders: Affinities of the family Penestomidae (NEW RANK), generic phylogeny of Eresidae, and asymmetric rates of change in spinning organ evolution (Araneae, Araneoidea, Entelegynae)", Molecular Phylogenetics and Evolution, 55 (3): 786–804,

doi:10.1016/j.ympev.2010.02.021 {{citation}}: Unknown parameter |lastauthoramp= ignored (|name-list-style= suggested) (help). At least half the authors are well-known in arachnology, and could easily end up with Wikipedia articles; the journal already has an article. Adding yet more links/IDs for every author and the journal just seems inappropriate to me, as opposed to a single link to a Wikidata entry for the citation. Peter coxhead (talk

) 13:19, 14 April 2016 (UTC)

How many of our citations name 15 authors? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:59, 14 April 2016 (UTC)

Andy, it depends on the area. In molecular phylogenetic studies, increasingly important in most organism articles, many authors are common; also in physics. Is it not possible to have a single data link for the citation? If I want to look up other works by book authors, for example, I use the OCLC WorldCat entry, like this one. I can only repeat that a single link to a data entry for the citation which then linked to authors, work, etc. would be much more likely to find support here than separate links for every author and work. As Jason also pointed out, it's a bad database design to have the amount of redundancy that would result if every article that used a citation had independent links to the authors and the work. Better to get the data design right first, surely? Peter coxhead (talk) 14:28, 14 April 2016 (UTC)

My question was how many of our citations have 15 authors. You'll also note that I started thissection by talking about "a work, or journal, which has... a Wikidata item". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:54, 14 April 2016 (UTC)

As an aside: I have no idea of just "how many", but I just did three citations that had 18 or 19 authors. Such lengthy lists of authors are often shortened to "Smith and 14 others", but the problem there is that the most significant author may put his name at the end. Or several notable authors may get buried in an alphabetical list. A reader trying to independently assess the quality of the authorship would be greatly assisted if the most notable authors had wikilinks. ~ J. Johnson (JJ) (talk) 19:42, 14 April 2016 (UTC)

Can you give me an example of where such a cross-project link benefits the reader? Resolute 13:54, 14 April 2016 (UTC)

Yes; see the Reasonator links in {{

Red Wikidata link}}. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits

13:59, 14 April 2016 (UTC)

The example in that Reasonator link (Bach) argues utility as an external or interwiki link for the subject of an article, but I am wondering about the utility of adding a Wikidata link inside a citation template, as is requested here. Particularly given you are framing this around an author without an article, and therefore presumably non-notable. Resolute 14:30, 14 April 2016 (UTC)

I didn't say anything about Bach. And no, I'm not "framing this around an author" of any kind; I asked about "a work, or journal [or] author". But notability is a red herring; we don't require authors or their works to be notable before citing them; nor do we require works with DOIs or ISBNs to be notable before linking those identifiers. And if there is a Wikidata item, then they meet notability requirements there. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:54, 14 April 2016 (UTC)

I still see don't see how this really benefits readers even after your template example. I think Resolute's question deserves an in-depth, well-presented and detailed reply. Your idea is an interesting possibility but how many readers would this really help? I suspect very few. As it stands, the idea seems like it belongs in the "solution in search of problem" category. Jason Quinn (talk) 16:36, 14 April 2016 (UTC)

As I noted above: Where we cite a work, or journal, which has a corresponding Wikipedia article, we can make the title a link; and we can use |authorlink= if there is an article about the author. Presumably, we are all agreed (or at least have consensus) that these links are useful to our readers, in a number of ways, which is why we provide them. Where there are no no such articles, but we have information in Wikidata, with a human-friendly interface available at Reasonator, linking to one, or the other (whether on the page or in metadata) will be useful in the same ways. I cannot see any good reason why we would deliberately not inform our readers and re-users that we have additional information on something or someone that is cited in an article. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:37, 14 April 2016 (UTC)

Sorry, but "it is valuable because it is" is not compelling to me. I just want to understand what utility your proposal offers the reader. What "additional info" can I expect Wikidata to provide me in this scenario? Resolute 20:53, 14 April 2016 (UTC)

@Resolute: My argument is not "it is valuable because it is"; it is "it is of comparable value to the links we already provide". Properties which Wikidata may offer, about an author for whom we have no Wikipedia article, include (but are not limited to):

• Alternative names/ name spellings/ translitterations
• Text description (multilingual)
• Affiliations (past and present)
• Alma mater
• Links to Wikipedia articles in other languages
• Unique identifiers (VIAF, ISNI, ORCID, ReseacrherID, and many others, some domain-specific)
  • These may link to other works; library catalogues, etc
• Website
• Social media links
• Image (and Commons category if multiple free images are available)
• Audio recording of voice, including preferred name pronunciation
• Nationality
• PhD supervisor
• PhD students

and of course disambiguation from other, similarly-named, authors. I'm short of time now, but can provide a similar list for works, later. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:26, 14 April 2016 (UTC)

I guess part of my problem is that I don't agree that it's normally useful to provide wikilinks to authors and works in citations, and I very rarely do so. They distract from the important link, which is to the source, and create duplicated wikilinks, which are hard to fix, since references often get changed and reordered: what may have at one time been the first occurrence of an author soon changes. On the other hand, I do absolutely agree that it's useful to provide access to the information Andy has listed above. I'd just like to find a way of doing it which doesn't potentially involve links to every author in every citation, which I don't think is the right way forward. Peter coxhead (talk) 10:11, 15 April 2016 (UTC)

The reliability of a source depends in large part – and its authority entirely – on the authors, which makes information about the authors (or at least the principal authors) of a source potentially useful informaton. (That is why I have proposed [above] providing for author links of some kind.) On whether links to Wikidata items would be good I have no opinion. But author links generally are potentially useful. This does not mean that "every author in every citation" should (or will) be linked. ~ J. Johnson (JJ) (talk) 20:58, 15 April 2016 (UTC)

Well, as I noted above, there is both consensus and common practice to do provide links, so any objections on that score are moot. And not all sources are online, of course. A ticket to make removing duplicated wikilinks in references part of the core MediaWiki could be raised, if anyone considers the issue sufficiently significant. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:23, 15 April 2016 (UTC)

You know, I'd be far more interested in supporting your ideas if you stopped treating everything you want as fait accompli. While you can argue that consensus exists to allow for links to Wikipedia articles for authors in the citation templates (though I too find them unnecessary and distracting), you do not have consensus for Wikidata links. And on that point, bluntly, I currently oppose their addition. Resolute 13:13, 15 April 2016 (UTC)

I'm not sure I can help you overcome the misapprehension that the question "How might we include, and display, the 'Q' value, and link to the item?"" is "treating everything [I] want as fait accompli". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:21, 15 April 2016 (UTC)

Oppose. We don't usually link to articles about sources; I think it's best to save that for cases when it is particularly relevant. Indeed, we don't even have a way to link to both provide an external link to the source itself and also to a Wikipedia article about the source, because linking to Wikipedia articles about sources has never been a priority. As for authors, linking to the author's Wikipedia article serves as a hint to the reader that the author is notable enough to not only qualify for a Wikipedia article, but induce some editor to go the effort of writing the article. If all authors have links to somewhere, we will just have a sea of blue. Jc3s5h (talk) 13:25, 15 April 2016 (UTC)

You oppose the asking of a question? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:21, 15 April 2016 (UTC)

I'm not in favor of this idea. If I want to know where a wikilink goes, I float the mouse cursor over the link and the tool tip shows the underlying link. So, for Douglas Adams, it shows 'Douglas Adams'. If I link to the same title at de.wiki, the tool tip shows 'de:Douglas Adams'. If I link to the same title at wikidata, the tool tip shows me 'd:q42'. To know what the 'd:q42' link actually points to, I must follow it. There is no way to know by simple inspection that 'd:q42' leads to something meaningful. Wikidata does have the concept of aliases but unless something has changed, doesn't support redirects of those aliases so the only way to guarantee that you will land on a particular q value is to use that value in a link. This is much too unfriendly to humans. Readers and editors should be able to identify the link destination without needing to follow the link.

—Trappist the monk (talk) 13:39, 15 April 2016 (UTC)

You seem to be hung up on the user interface that you experience; others have a different experience (and we can configure title attributes on elements, to generate preferred tooltips). Nonetheless, my question about "How might we include, and display, the 'Q' value, and link to the item?" does not presuppose any particular interface. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:21, 15 April 2016 (UTC)

Yeah, I am hung up on the user interface. Constructs that generate preferred tooltips like this are not a solution, just a mask:

[[d:q42|<span title="Douglas Adams">Douglas Adams</span>]]

Douglas Adams

If the link is wrong, [[d:q24|...]] instead of [[d:q42|...]], a common typo or digit transposition:

Douglas Adams

the reader still doesn't know where the link leads without following it because the title attribute masks the usual MediaWiki tooltip (which for WikiData is meaningless to humans anyway). Also, because there is no redlink facility for interwiki links:

[[d:42|<span title="Douglas Adams">Douglas Adams</span>]]

Douglas Adams

both readers and editors don't know that the WikiData link is malformed.

The parser function {{ifexist:}} is no help here because it does not work with interwiki links.

—Trappist the monk (talk) 10:26, 16 April 2016 (UTC)

It's possible, in Lua, to obtain the text label of the Wikidata item, which could then be used as the tooltip. A transposed digit would then be indicated by that text. In any case, we don't avoid "building the web", as User:RexxS put it, just because it's hypothetically possible to use the wrong link. Similarly, malformed Wikidata values can be trapped by error detection in Lua. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:49, 16 April 2016 (UTC)

It is: mw.wikibase.label( 'Q42') returns 'Douglas Adams'.

—Trappist the monk (talk) 12:39, 16 April 2016 (UTC)

Imo, "building the web" is a newfangled idea when it comes to citation templates. They exist for a specific purpose, i.e. as standardized presentation tools for citations, which in turn are required by policy. It is not the business of cs1 to build the web; this is best done in properly cited article text. 65.88.88.71 (talk) 19:26, 16 April 2016 (UTC)

Oppose. Turning a reference list into a sea of blue is a really bad idea. Including a large number of author links will obscure more useful links such as those to the original source. Author links should be reserved for notable authors for which an article has been written. Even in these cases, I question their usefulness. While the author of the source may be important in determining reliability, the type of source (primary vs. secondary) and where it was published (reputation for fact-checking and accuracy) are probably more important. Including gratuitous author links amounts to little more than hyping the author. Boghog (talk) 09:33, 16 April 2016 (UTC)

You oppose the asking of a question? No-one has suggested "a sea of blue". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:49, 16 April 2016 (UTC)

Obviously you would not have asked the question if you were not proposing to implement it in some form. Since I do not see how this could be implemented without facilitating the addition of low value links in citations, I oppose the implementation of the suggestion. Using |author-link= is far more appropriate since the target of these links have already passed a notability threshold. Boghog (talk) 13:04, 16 April 2016 (UTC

You may claim it is "obvious"; but it is not so. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:29, 17 April 2016 (UTC)

Oppose those links. Andy may just technically be "asking questions" but there's clearly a desire to work wikidata links into the cite templates and therefore an implicit proposal (that can be easily extrapolated from the framing of the discussion) to allow them there. I don't think they belong in the cite templates so I would oppose such links. But to answer your question, Andy, there's only two ways to display the Q-ids that seem to merit attention: next to the author name or as a tooltip over the author's name. Putting them elsewhere (like bunched in with the doi, isbn, and such) would be too ugly for real consideration. Above you argue that since we wikilink authors' articles, you see no reason why we shouldn't wikilink to identifiers, since they both "will be useful in the same ways". This is a false analogy and ignores other salient aspects of the idea like usability (as I already mentioned). This argument of yours was generalized into a "Build the web" argument, which I think is particularly bad. Wikilinks are best when chosen sensibly, not just because a relevant link does exist. What's next linking to author's listing at Wikiquote? Or the publisher location to Wikidata and Wikivoyage? "Build the web" is an easily abused idea and invoking it here reminds me of

WP:ALLWIKI. Jason Quinn (talk

) 23:52, 16 April 2016 (UTC)

The "usability" FUD is easily addressed by using Reasonator links, as I mentioned some time ago in this thread. As to your reducto ad absurdum, if an author has a Wikiquote link, or a publisher's location has a Wikivoyage page, then these will be available in Wikidata, and thus in Reasonator. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:29, 17 April 2016 (UTC)

Support and potential compromise solution. I support Andy's (and to some extent also JJ's) idea in general. Citations are used to verify the information in an article but are also used as a starting point for further research into a subject - and that's not a by-product, but part of the whole idea of why citations are provided in the first place. The author info is an important part of that and if we have a chance to reliably identify an author beyond his/her name, we should use it.

While I could accept the |author-id= idea proposed by JJ, like others I'm not too fond of the idea of potentially having half the alphabet of superscript links following an author's name in the end. Ideally, a single link should be sufficient, and we already have the |author-link= parameter to provide such a link, it just needs some further tweaking to improve the appearance.

Based on the original proposals and various pro and contra comments further above, my compromise proposal would be as follows:

If we have an article about an author, we link to it through |author-link=. If we lack an article in the English Wikipedia, but have one in another Wikipedia, we link to that article (via prefix code) until someone has created an article about the corresponding author in the English Wikipedia as well. In both of these cases, all the various authority ID codes (including links to Wikidata) are added to the article about the author (indirectly, this will end up at Wikidata via bots). In case we don't have an article about an author in any Wikipedia, we can use |author-link= to link to a Wikidata entry instead (if it exists). In this case, the various authority ID codes can be added to the Wikidata entry. As has been pointed out already, links to Wikidata are a bit cryptic, and the Wikidata appearance isn't particularly friendly to readers. However, as was discussed further above, it would be possible for the citation template to pull data from Wikidata in order to improve the local appearance of citations. But instead of directly linking to Wikidata wouldn't it be possible for the citation template to link to some kind of special "author page" as a frontend for Wikidata similar to what we do for ISBNs linking to Special:BookSources/????????????? (providing the Wikidata knode as a parameter) and then have that page pull the various data from the actual Wikidata entry? This page would only be invoked if a reader clicks on the author link, so it would also address the problem that pulling data from Wikidata is "expensive". Of course, this indirection should only happen, if the link provided by |author-link= starts with "d:" (for Wikidata).

Further, whenever the citation template detects that |author-link= links are starting with a prefix (other than "en:"), it could append the blue "up-arrow" symbol (normally used to indicate external links) to the linked author's name. This would serve as a visual clue for readers that following the link will leave the English Wikipedia (but still point to a sister project, that is, not leave the "Wikipedia universe" in general (by that I count Wikidata as part of the Wikipedia infrastructure)). (Such links were still easily distinguishable from true external links provided via urls, which are shown in [square brackets] (and cause an error message, anyway). If this wouldn't be enough, we could use a different color for links to sister projects.)

I think, this compromise proposal would provide means to reliably identify and distinguish between authors of the same name and to aid further research into a subject as well as help to further build the web in general, while avoiding the potential clutter and confusion caused by multiple author-related links in a citation. At the same time it would maintain the principle of minimal surprise for readers and minimize the processing costs for machines. (The proposal does not provide a direct solution in case that even a Wikidata entry does not exist yet, but if the author is important and needs to be distinguished from other authors of the same name, a Wikidata entry could be created at least by the more experienced users.)

--Matthiaspaul (talk) 17:09, 1 May 2016 (UTC)

Building the web

The relevant advice for me is

Fo me the issue is more about how to make the link to Wikidata. Ideally, I'd prefer citations to be stored in Wikidata, with the ability to generate differently formatted references here – long term I'm sure this will happen. The alternative is presumably something that works like {{

Interlanguage link multi}}, i.e. it generates a link to the Wikipedia article if it exists and to Wikidata otherwise. This does have to use the resource-expensive call to #ifexist, which is considered undesirable – but maybe this has changed? If not, I suspect it would not be a good idea to have it being called from huge numbers of citations. Peter coxhead (talk

) 16:51, 15 April 2016 (UTC)

@RexxS: Who is denying the ability to make a link? Just use |authorlink=d:... as 72.43.99.130 already said early in this thread. I am still not convinced that doing so is a good idea, but the ability is there. No change to the software is needed. —David Eppstein (talk) 17:09, 15 April 2016 (UTC)

As noted above, that does not indicate that the target is Wikidata, not Wikipedia. See also Jason Quinn's comments on that suggestion, above. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:21, 15 April 2016 (UTC)

@Peter coxhead: Re prefering citations to be stored in Wikidata; that's what I envisage too, and have done for at least a couple of years; see User:Pigsonthewing/Citations - the future. I'm just asking, here, how we might store and. or display relevant metadata in the interim. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:38, 15 April 2016 (UTC)

And my answer above was that we could explore, as an interim solution, whether something that works like {{

Interlanguage link multi}} would be acceptable. Can you comment on the resource cost of all the #ifexist calls? Peter coxhead (talk

) 06:26, 16 April 2016 (UTC)

I'd suggest looking at coding a Lua module that takes the author's name and returns either: the name enclosed by a link to the wikidata entry, where the entry exists; or the bare name, where the entry does not exist. That would give a proof-of-concept for the required functionality, although incorporating it into the CS1 citation module would be a little more complex, depending on whether another parameter would be allowed or auto-detection would be required given the present parameters. There is no need for #ifexist because the mw.wikibase calls return nil if the entity does not exist, and that is easy to test for. Unfortunately the use of arbitrary access in the mw.wikibase call is also an expensive function, so your concern remains valid. We couldn't judge the impact, however, without some trials. Hope that helps. --RexxS (talk) 13:19, 16 April 2016 (UTC)

Automatic linking of any kind based on the author's name is wrong. If this idea is pursued (which I don't favor) It would be the responsibility of the human editor who adds the citation to determine whether any Wikidata item to be linked to actually refers to the same person who wrote the source being cited. A mere name match is not sufficient. Jc3s5h (talk) 13:28, 16 April 2016 (UTC)

The biggest problem with this proposal is the general lack of notability and relevance of the linked authors. The reason a source is cited is not because of who wrote it, but because the source is reliable and supports the statement that is being made. In cases where the author does matter, |author-link= is available. If and when Wikidata does start storing citations and associated author data, the automatic generation of Wiki data links will turn the reference section of articles into a sea of blue. Better to provide a single link from the source to Wikidata. If someone really is interested in a particular author, they can link to the author from Wikidata. Boghog (talk) 17:28, 16 April 2016 (UTC)

Two points: the present |authorlink= automatically links to the Wikipedia article, so how can "automatic linking of any kind" be wrong? If it's good enough for one project, it's good enough for its sisters. It's already the responsibility of the contributing editor to ensure that the author linked to the Wikipedia entry is the same author as the source, so why does this somehow become any more of a problem if the target is Wikidata?

One important reason why we may want to link to an author is when the reliability of the source depends on the author being an acknowledged expert in the relevant field (

WP:IRS #Exceptions). It is surely a worthwhile convenience for any editor checking the verifiability of the text to have a link to an author in such circumstances. This would be especially true in specialised fields where the author may only be well-known in a foreign country and only have a Wikidata entry. It's not an overwhelming reason for providing Wikidata links, but is sufficiently relevant to counterbalance concerns over any lack of usefulness.— Preceding unsigned comment added by RexxS (talk • contribs

) 17:52, 16 April 2016 (UTC)

@RexxS: your proposal is not detailed, but appears to suggest that if the editor provides the name(s) of the author(s) with last1 =, first1 =... then the template will automatically try to find one of two kinds of links, either within Wikipedia or within Wikidata. This is unacceptable. It is the duty of the editor who adds the citation (if he/she desires a link to the author's information) to manually find a suitable entry in Wikidata or Wikipedia, and code the template to specifically indicate which entry is the one that describes the author(s) of the source. Jc3s5h (talk) 18:07, 16 April 2016 (UTC)

It may be unacceptable to you, but it's not to me or many others. I have never suggested that the template/module should use |first= and |last=, as that would obviously make disambiguation impossible. I was proposing that either (1) the name supplied at |authorlink= might first check for a Wikipedia article, and in its absence check for a Wikidata entry; or (2) a new parameter might be used to supply a Wikidata link. Either of those two schemes would be feasible; but the former would be more costly, albeit easier to use, whereas the latter would be more robust and easier to code, but would require more effort on the user's part. Frankly, it's not helpful in trying to explore possibilities when you make a blanket opposition with no better rationale than your declaration that it's the editor's duty to do such-and-such. If you had your way, we'd still be crafting citations by hand. --RexxS (talk) 23:05, 16 April 2016 (UTC)

A name supplied at |author-link= is likely to encounter problems finding the right person at Wikidata. For example:

John Smith (explorer) – mw.wikibase.label('Q228024') returns John Smith

John Smith (architect) – mw.wikibase.label('Q1243465') returns John Smith

which brings us back to the very unfriendly |author-link=d:Q228024 ...

—Trappist the monk (talk) 00:17, 17 April 2016 (UTC)

Yes, I quite agree that it's not easy trying to find the correct entry when Wikidata labels are not unique, but the sitelinks are. So mw.wikibase.getEntityObject('Q228024').labels[mw.language.getContentLanguage().code].value will return "John Smith (explorer)"; similarly for 'Q1243465'. It's also possible to deal with other language articles by replacing mw.language.getContentLanguage().code with a specific language code. The mw.wikibase.sitelink('QQ228024') call also performs that function, but without the ability to refer to other language Wikipedias. Once we have the QID that the supplied parameter refers to, we know which Wikidata article is intended and we can supply a disambiguated tooltip or text for the link. There would still be effort needed on the part of the contributing editor to supply a properly disambiguated value of the |authorlink= parameter, of course, but that same diligence is already required when we are linking to John Smith's article on Wikipedia. Anyway, I'm just throwing ideas around. It may be that there's insufficient demand to make the programming worthwhile, but I don't think it hurts to explore the possibilities and difficulties involved in adding functionality. Cheers --RexxS (talk) 20:41, 17 April 2016 (UTC)

According to the debug console, mw.wikibase.getEntityObject('Q228024').labels[mw.language.getContentLanguage().code].value returns "John Smith" without the disambiguator which, upon inspection, makes sense: it is just a more verbose (and expensive) version of mw.wikibase.label('Q228024'). To get what it appears is really wanted, for the local wiki this is marginally better:

mw.wikibase.getEntityObject('Q228024'):getSitelink()

no need that I can see to do this unless we need to know what the local wiki is (but in that case we'd have called mw.language.getContentLanguage() elsewhere so this seems like overkill:

mw.wikibase.getEntityObject('Q228024'):getSitelink(mw.language.getContentLanguage().code .. 'wiki')

Modifying that and just passing the 'name' of the wiki you want the sitelink of: 'frwiki', 'ruwiki', 'nowiki', works too:

mw.wikibase.getEntity('Q228024'):getSitelink('ruwiki') returns: Смит, Джон (капитан) – Smith, John (Captain) according to Google translate

But, the documentation says that mw.wikibase.getEntity() is expensive when not related to the current page. If John Smith (explorer) is an author, calling mw.wikibase.getEntity() to find the sitelink in another wiki will bump the expensive parser function counter. That's the toll that the parser troll will exact from you if you want to look outside the local wiki. If you don't need to look outside, perhaps use mw.wikibase.sitelink() is best.

—Trappist the monk (talk) 23:27, 17 April 2016 (UTC)

Thanks for checking that call and finding the correct one. Yes, a module will normally have created a variable to hold mw.language.getContentLanguage().code - have a look at Module:Wikidata, for example. I agree that mw.wikibase.sitelink() is simplest if you only need to find the label for the local wiki. As I said above to Peter coxhead, as soon as you try to access a Wkidata entry other than the one corresponding to the current page (called "arbitrary access"), you make an expensive call. That means every author that we might want to link to Wikidata would incur that penalty. I don't know how much overhead that is (say, compared to expensive parser functions), and we'd have to do tests to ascertain what kind of limit it might impose on the number of citations that would be allowed. It may be that such a limit would make this suggestion impractical, but I still think it's worth exploring these sort of possibilities for added functionality to our templates. --RexxS (talk) 08:58, 18 April 2016 (UTC)

[The] present |authorlink= [does not] automatically [link] to the Wikipedia article. An editor must supply an article title.

—Trappist the monk (talk) 18:12, 16 April 2016 (UTC)

Yes on why we would want a link to an author. But I would concur with Boghog and TM that such links should not be automatic, but contingent on an editor finding and evaluating their appropriateness. And I am not so confident on the general adequacy of Wikidata items. But wouldn't links to ORCID, etc., suffice for the purposes that have been mentoned here? ~ J. Johnson (JJ) (talk) 21:01, 16 April 2016 (UTC)

@Trappist the monk: The present |authorlink=Douglas Adams does automatically link to the article Douglas Adams. Obviously the author's name must be supplied, otherwise disambiguation would be impossible. So I'll ask you again: (1) how can "automatic linking of any kind" be wrong? (2) It's already the responsibility of the contributing editor to ensure that the author linked to the Wikipedia entry is the same author as the source, so why does this somehow become any more of a problem if the target is the author's Wikidata entry? --RexxS (talk) 22:51, 16 April 2016 (UTC)

Please do not put words into my mouth that I have not spoken. I have said nothing regarding the rightness of wrongness of 'automatic linking'. All I said is that the notion that |author-link= is automatic is not true. To me, an automatic link happens without editor intervention. Such a link might result when the module reads |author= or a |last=/|first= pair and decides on its own where to link that author name. Clearly this does not happen.

—Trappist the monk (talk) 00:17, 17 April 2016 (UTC)

My sincere apologies. The comment that "automatic linking of any kind" was made by Jc3s5h, not you. I was referring to the linking that is made when plain text is supplied to the |authorlink=Some Name parameter. In the context of the current discussion, it is pertinent that a link is automatically created to a Wikipedia article with that text, e.g. Some Name. By analogy to your reasoning above, it is clear to me that the module reads the value of |authorlink= and decides on its own to create an internal link without editor intervention. Trivial as it seems, clearly that does happen. It is that automatic behaviour that this very discussion is examining, by suggesting that perhaps it might not create that internal link automatically if it would be a redlink, but would search for another usable link from the plain text supplied. It seems you require a greater degree of complexity in your definition of "automatic" than I do. That's OK, I understand your usage now. --RexxS (talk) 21:01, 17 April 2016 (UTC)

RexxS, that's not an automatic link - somebody needs to fill it in, so it's manual. If it were automatic, the link would be made even if the |authorlink= param were blank or omitted. --Redrose64 (talk) 23:00, 16 April 2016 (UTC)

If it were not automatically a link, we'd have to supply |authorlink=[[Douglas Adams]], rather than |authorlink=Douglas Adams. I can see we're thinking about different meanings of the word automatic, but in the context of Jc3s5h's "Automatic linking of any kind based on the author's name is wrong", either of our interpretations refutes his assertion. --RexxS (talk) 23:13, 16 April 2016 (UTC)

I think that some people - myself included - are remembering a discussion a few weeks ago where the proposal was that if |author-link=y were present, the citation template would link based upon |author= or |first=|last=. This is what I was understanding by "automatic". --Redrose64 (talk) 08:56, 17 April 2016 (UTC)

OK - I understand now. I hadn't seen that discussion, but it must be immediately obvious that you can't create reliable links of any kind from |author= or |first=|last= because of the impossibility of disambiguation in that scheme, since adding disambiguators to an author's name will screw up the metadata for third-party re-users. --RexxS (talk) 20:41, 17 April 2016 (UTC)

Alternative implementation

I point out that the discussion here about linking authors to wikidata is entirely in the context of use within |author-link=, and particularly as an alternative where there is no article for a regular wikilink. I further point out that this is not the only possible context. E.g., the |author-id= parameter I am proposing (above) could accommodate wikidata links, independently of of |author-link=s. I don't know if that would be a good idea, but it could be done. Indeed, such linkge could be readily added to my {{authorid}} template. If wikidata links are a good idea, this could be an easy way to implement them. ~ J. Johnson (JJ) (talk) 21:49, 18 April 2016 (UTC)

How is it "entirely in the context of use" of |author-link=? If you mean more than "technically possible", then I would disagree. Help:Citation Style 1 indicates not once but twice that |author-link= should be used for the name of the Wikipedia article about the author. It also says not to use |author-link= to link to external websites and I think it's open for debate whether cross-project links count as external or not here. I think they should count as external links. The non-WMF interwiki links like this one to IMDB clearly should count as an external link. It is not so obvious for WMF-run but cross project links like those to Wikidata. {{Citation Style documentation}} says basically the same as Help:Citation Style 1 so it is a firm conclusion that the documentation says these links are non-standard and contrary to the existing cite template documentation simply because the Wikidata links are not articles but possibly also because they should count as external links. Jason Quinn (talk) 16:34, 19 April 2016 (UTC)

You have confused the descriptive with the proscriptive. What I said is that the discussion here is (description) "entirely in the context" of using author-links where there is (to quote the title of this topic) "... a Wikidata item, but no article". What you are saying is that |author-link= should (proscription) not be used for this purpose. Which I find persuasive, but you should have raised those points above. What I am suggesting here is an alternative that does not use |author-link=. An alternative entirely conformable with your comments. ~ J. Johnson (JJ) (talk) 17:16, 19 April 2016 (UTC)

I've seen several references where an "|access-date= is missing a URL" error is thrown, however a DOI is present. I've been resolving this error by following the DOI, and using the resultant URL for the |url= parameter. But isn't the point of DOI to maintain links that may change? Can we make |access-date= accept |doi= as an acceptable URL? If so, I'd be interested in helping, to learn how this change would work. - Paul2520 (talk) 23:42, 1 May 2016 (UTC)

It seems I posted this too soon. I was able to find this archived discussion, which discusses reasons against doing what I suggest. - Paul2520 (talk) 23:44, 1 May 2016 (UTC)

@

books

} 00:39, 2 May 2016 (UTC)

Is there a way, or could a way be made, to track population of the via= parameter: what articles populate it, when was it added, who added it, etc.? Tks 

Perhaps. Certainly we could create a category to hold pages that have cs1|2 templates using |via=. As for when [it was] added, who added it, etc. I don't think so. How would this category be organized? By page name or by the name of the value assigned to |via=?

—Trappist the monk (talk) 11:50, 1 May 2016 (UTC)

"By value assigned" sounds more useful, so we could get forex a set of subcats or something, e.g. "[Category:Pages via Highbeam]" or....? And then, I know this may be too much, but just brainstorming, a bot could check the subcats and monitor changes? I understand the category/subcats probably couldn't track who did what when, but perhaps a bot could... I dunno if it would be worth the trouble. Could try and see..? Thanks for your reply.  

Lingzhi ♦ (talk)

13:06, 1 May 2016 (UTC)

I was thinking of something more general as a starting point simply because there is no limit on what editors can place in |via= (there are nearly 18k pages with |via= in some form or other) so:

for |via=Highbeam and |via=Highbeam Research, Module:Citation/CS1 would add the category [[Category:CS1 templates using via|H]]

for |via=LexisNexis: [[Category:CS1 templates using via|L]]

where the sort keys would control how the articles are listed in the category.

Later, if it makes sense, we could create a list of those specific sources that the module should place in subcategories of:

[[Category:CS1 templates using via]]

[[Category:CS1 templates using via Highbeam]]

[[Category:CS1 templates using via LexisNexis]]

—Trappist the monk (talk) 15:56, 1 May 2016 (UTC)

Why do we even have a via parameter in the first place? It's completely useless and only introduce clutter in the reference section. What does it matter if a source was accessed through Highbeam or JSTOR, or ScienceDirect or LexisNexis?

books

} 18:02, 1 May 2016 (UTC)

When we brought |subscription= into cs1|2 as a replacement for {{

here

.

Apparently there was once some documentation at Wikipedia:The Wikipedia Library that discussed the need for |via= but if its still there I didn't find it. Something to do with attribution I think.

—Trappist the monk (talk) 19:25, 1 May 2016 (UTC)

Even without |via= there are users who insist on including information on where they found the source. --Redrose64 (talk) 19:43, 1 May 2016 (UTC)

Which is pointless for journals and other published sources of that nature, since they are identical across all methods of access.

books

} 19:58, 1 May 2016 (UTC)

You think I didn't try telling them? Have a look at the page history as well, see how many times they (or an IP) put it back in after I removed it. --Redrose64 (talk) 21:05, 1 May 2016 (UTC)

This is a good illustration of the principle that if we supply a parameter, editors will believe that it should be used, and insist on using it. So we should be careful what we supply. —David Eppstein (talk) 21:10, 1 May 2016 (UTC)

"Via" may be necessary per Wikipedia's own policies. Contributors must say where they got the supporting material, so the particular source can be verified. There is no such thing as a "permanent" online link. The most "stable" of identifiers can be re-assigned, go stale, or just disappear. The citation will then be incorrect, as the orginally accessed version of the source is no longer accessible. Also, for the purposes of Wikipedia, where everything is done anonymously (I've no idea what "David Eppstein" means – no offense meant, this is only for illustration), we cannot just assume that any and all online re-publishers are reliable, or reliably accurate. At least "via" offers the reader an indication that what is cited is an online reprint by another publisher, and not the original publisher's version. As stated below, "via" has nothing to do with access requirements, but with the source's provenance, per doc. 65.88.88.126 (talk) 13:12, 2 May 2016 (UTC)

The use of |via= is suggested by the examples at each of the TWL partner pages, e.g. Wikipedia:HighBeam. Wikipedia:The Wikipedia Library/Partners lists "Publisher credit using the |via parameter of our citation templates" among the benefits to publishers, though this is at odds with the later statement on the same page that the partnership is not "An agreement to advertise the resource services beyond what is normally done for the use of any source". Kanguole 05:23, 3 May 2016 (UTC)