IRC office hours/Office hours 2013-04-25

From Meta, a Wikimedia project coordination wiki

[19:02:54] <Denny_WMDE1> hi — anyone here for the extra office hour on references in Wikidata?
[19:04:52] * marktraceur has changed topic for #wikimedia-office to: " ./phpunit.php --group Parser"
[19:04:56] <marktraceur> Damn it
[19:05:02] * marktraceur has changed topic for #wikimedia-office to: "answering questions about the sources system on Wikidata to help with the references and sources RfC"
[19:05:45] * marktraceur has changed topic for #wikimedia-office to: "answering questions about the sources system on Wikidata to help with the references and sources RfC | | Wikimedia Office Hours:"
[19:06:49] <Denny_WMDE1> marktraceur: wot?
[19:07:33] <marktraceur> I messed up the topic by failing at pasting
[19:07:45] <Denny_WMDE1> :)
[19:07:53] <legoktm> oh when does that start?
[19:07:59] <marktraceur> 7 minutes ago
[19:08:07] <legoktm> erp
[19:08:16] <Denny_WMDE1> hi all
[19:08:19] <Sven_Manguard> what office hour is it?
[19:08:27] <Denny_WMDE1> about sources and references in wikidata
[19:08:33] <Micru> hi Denny_WMDE1
[19:08:39] <Denny_WMDE1> small extra office hour, just to discuss stuff if needed
[19:08:44] <Sven_Manguard> Ah, that one that I totally knew about before seven seconds ago
[19:08:52] <Denny_WMDE1> :)
[19:09:11] <Denny_WMDE1> to be honest, it seems that most of it is pretty figured out already
[19:09:28] <Denny_WMDE1> the URL datatype is sorely missing, and a few other datatypes too
[19:09:53] <Micru> probably the things left to discuss are 1) language and references 2) url and 3) reference annotations
[19:10:24] <Denny_WMDE1> so in general it will be possible to query for data of a reference in a wikipedia page
[19:10:42] <Denny_WMDE1> this means, there is no need to repeat the author in every reference
[19:11:04] <Denny_WMDE1> ah, wait a second — do we all agree on the usage of the terms "reference" and "source"?
[19:11:15] <Denny_WMDE1> so source for me is an item in wikidata
[19:11:21] <Micru> well, let's clarify that first :)
[19:11:32] <Denny_WMDE1> a reference for me is the list of property-key values connected to a statement
[19:11:39] <Micru> ok
[19:11:43] <Denny_WMDE1> most often, the first one will point to a source
[19:11:46] <Denny_WMDE1> i.e. an item
[19:12:10] <Denny_WMDE1> does this make sense as a working terminology for us for now?
[19:12:15] <Denny_WMDE1> it is not perfect, i know
[19:12:38] <Micru> yes, for me that is clear, but in the RFC there were some missunderstandings
[19:12:46] <Micru> maybe a clarification in the glossary will be needed
[19:12:52] <Denny_WMDE1> yep, i guess we only worked it out on the go
[19:12:58] <Denny_WMDE1> i wasn't so clear about this myself
[19:13:06] <Denny_WMDE1> and need to think about it consciously not to mess it up
[19:13:08] <Denny_WMDE1> but it makes sense
[19:13:20] <Micru> in wikipedia they use other terms, so it makes it confusing
[19:13:35] <Micru> "notes" for references and "references" for sources
[19:13:59] <Denny_WMDE1> should we switch to that?
[19:14:09] <Denny_WMDE1> i have no preference on terminology
[19:14:20] <Micru> I wouldn't mind
[19:14:48] <Denny_WMDE1> anyone has a preference? :)
[19:15:11] <Micru> don't worry, as soon as we post it, someone will complain, it is the rule :)
[19:15:15] <Denny_WMDE1> hehe
[19:15:41] <Micru> anyway, what i meant before is that maybe some statements have different references depending on the language, or different notes for the same item
[19:16:14] <Micru> so maybe it is needed some kind of language preference for statement sources
[19:16:37] <Denny_WMDE1> Micru: i don't know
[19:16:52] <Denny_WMDE1> it is a a refernece for the statement
[19:16:52] <Micru> it depends on how many, maybe it won't be an issue
[19:17:03] <Denny_WMDE1> nomatter what the language
[19:17:26] <Micru> if there are 6 references, how do you chose which one would display in wikipedia?
[19:17:31] <Denny_WMDE1> all?
[19:17:35] <Micru> ah :)
[19:17:40] <Micru> that makes sense
[19:18:12] <legoktm> ok i had a question
[19:18:33] <Denny_WMDE1> it could be maybe a bit smarter than the [1][2][3][4][5][6] thing that we have now, but that's something we'll figure out on our way, i guess
[19:18:36] <Denny_WMDE1> legoktm: go ahead
[19:18:45] <legoktm> so on wikipedias we have citation templates say for a book, which would require you to add the title and the page #
[19:18:51] <Denny_WMDE1> by the way, your suggestion for the commonscat template is awesome!
[19:18:54] <legoktm> :D
[19:19:08] <Denny_WMDE1> it completely blew my mind today :) thanks for that
[19:19:11] <Denny_WMDE1> yes
[19:19:18] <Micru> which suggestion?
[19:19:21] <legoktm> but on wikidata, you could just add the page title, and there would be no "Error: you didnt specify what page #" or something
[19:19:43] <Denny_WMDE1> Micru: this
[19:19:50] <legoktm> and i guess that falls in with the freedom blog post you? (or someone) wrote a while back
[19:20:02] <Denny_WMDE1> yes, it does
[19:20:12] <Micru> good idea!
[19:20:24] <legoktm> but i'd like to see something, maybe a gadget, that would say "citing a book?" and would recommend the right properties for you (title, isbn, page #, etc)
[19:20:27] <Denny_WMDE1> well, just because the software doesn't enforce it does not mean the community should not enforce it
[19:20:34] <Denny_WMDE1> legoktm: absolutely, i agree with that
[19:21:17] <Denny_WMDE1> legoktm: what i meant with my blogpost is merely that these are questions that the developers in berlin should not solve in a hard way. but if you want to put a template over it via javascript, heck go for it!
[19:21:24] <Denny_WMDE1> that's exactly what i would expect
[19:21:26] <Micru> another thing is which properties (that later on will show as "notes" in wikipedia) will be allowed to attach to references, all?
[19:22:59] <Denny_WMDE1> yes, all
[19:23:18] <legoktm> oh hm one more thing
[19:23:31] <legoktm> most citation templates have a |accessdate= parameter
[19:23:43] <legoktm> which says which date the source was used
[19:24:19] <legoktm> is this something that could possibly done in the software? automatically adding the date at which point the source was added?
[19:25:07] <DanielK_WMDE__> legoktm: possibly, yes, but i don't think that would be a good idea.
[19:25:08] <Denny_WMDE1> accessdate is when the external source was checked, no?
[19:25:17] <DanielK_WMDE__> having a bit of javascript for that though would be fine
[19:25:17] <Denny_WMDE1> not when the reference was added
[19:25:40] <DanielK_WMDE__> Denny_WMDE1: you are right, of course :)
[19:25:42] <Denny_WMDE1> actually it would be nice if there was an external bot that keeps track of those files, and checks them periodically
[19:25:52] <DanielK_WMDE__> could still be helpful to get today's date suggested, and be able to change it
[19:26:26] <legoktm> yeah
[19:26:35] <Denny_WMDE1> i have a question, if that one's answered
[19:26:38] <Micru> Denny_WMDE1: since the note won't be possible, what about getting the item creation date of the reference?
[19:26:55] <legoktm> it would be nice if the date datavalue would autosuggest times like "now" or "today"
[19:27:33] <Micru> legoktm, do you mean when adding properties to a item used as a reference/source?
[19:27:52] <legoktm> yeah
[19:28:06] <Denny_WMDE1> Micru: I didnt understand your question, sorry
[19:28:20] <Micru> ok, i rephrase it
[19:30:05] <Micru> we have an item that will be used as a source/reference for an statement, that item would display in wikipedia in the citation template and the parameter "accessdate" is needed, wouldnt it be the creation date of the item used as source?
[19:31:40] <Denny_WMDE1> no
[19:31:50] <Denny_WMDE1> the accessdate is not the creation date of the item
[19:32:00] <Denny_WMDE1> the accessdate is when the source was accessed
[19:32:06] <Denny_WMDE1> or am i misunderstanding?
[19:32:32] <DanielK_WMDE__> accessdate is mostly useful for urls
[19:32:42] <DanielK_WMDE__> for books etc, it seems pretty pointless
[19:32:45] <Sven_Manguard> Accessdate makes no sense on Wikidata
[19:32:49] <Denny_WMDE1> in which case it would not be in the source, but in the reference
[19:32:51] <Micru> "The accessdate is when you fetched the reference; the date is when the article was published. The url is the line like http://www.etc; copy and paste the url in if available."
[19:32:57] <DanielK_WMDE__> it's basically a stopgap for a version identifier
[19:33:04] <Sven_Manguard> exactly
[19:33:06] <Micru> interesting
[19:33:34] <DanielK_WMDE__> Sven_Manguard: sure it makes sense, when giving a url as the source of some statement. then it's useful to give an accessdate for the url
[19:34:05] <Sven_Manguard> perhaps as a qualifier, but...
[19:34:14] <Ironholds> that's pretty useful, in my experience
[19:34:31] <Ironholds> Links die and the content they point to change - having a reference of "it said X on Y date" makes it a lot easier to verify.
[19:34:51] <Ironholds> rather than just "It said X". No it doesn't, that's a 404! Or says something completely different now!
[19:34:52] <Micru> in that case a suggestion of "acccesdate=now" would be useful for links
[19:35:36] <Denny_WMDE1> then the reference would be first the url
[19:35:40] <Denny_WMDE1> and then the accessdate
[19:35:51] <Denny_WMDE1> there would be no source item involved
[19:35:56] <Denny_WMDE1> it would all be in the reference
[19:36:02] <legoktm> Sven_Manguard: well not a qualifier, just as one of the source statements
[19:36:13] <legoktm> "Accessed on" --> "datevalue"
[19:36:48] <Denny_WMDE1> yep
[19:36:56] <Micru> Denny_WMDE1, I think there are two kinds of references, one for weblinks (without item), and another kind for non-web materials (with item)
[19:37:05] <Denny_WMDE1> Micru: yep
[19:37:12] <Denny_WMDE1> does non-web material have an access date?
[19:37:26] <Denny_WMDE1> i mean, does it say "Genesis 4,2, lastaccessed=now" ?
[19:37:34] <Sven_Manguard> Denny_WMDE1: as far as I know, it does not
[19:37:39] <Denny_WMDE1> phew
[19:37:40] <legoktm> er, i think it does
[19:37:42] -*- legoktm checks
[19:37:44] <Sven_Manguard> what?
[19:37:47] <Denny_WMDE1> woot? that's funny :)
[19:37:51] <Sven_Manguard> but they're static
[19:37:52] <legoktm>
[19:37:57] <legoktm> there's an accessdate parameter
[19:38:05] <Sven_Manguard> that makes no sense
[19:38:21] <legoktm> We're Wikipedians! It's not supposed to make sense!
[19:38:25] <Micru> the access date is more for journals that are available online
[19:38:28] <Sven_Manguard> if you have the name of the book and the edition, the book isn't going to bloody change in 40 years
[19:38:31] <Denny_WMDE1> accessdate has a prerequesite a url
[19:38:44] <Micru> yes, if it has an url, it needs access date
[19:38:53] <Denny_WMDE1> accessdate on cite book has as a prerequisite a url
[19:38:59] <legoktm> ah, you're right
[19:39:02] <legoktm> i missed that
[19:39:42] <Micru> i think the url is optional, probably for verification purposes, that the book exists
[19:40:11] <Sven_Manguard> agreed, but I think that if we're using a URL as a source, the accessdate needs to be attached to it. We want to make sure that the accessdate is *solidly bound* to that URL and can't get mixed up with other URLs
[19:40:16] <Sven_Manguard> that's why I said qualifier
[19:40:43] <Denny_WMDE1> Sven_Manguard: i agree with your statement, just technically it is not a qualifier, but that's technical nitpicking
[19:40:47] <Sven_Manguard> Oh
[19:40:49] <Sven_Manguard> well
[19:40:53] <Denny_WMDE1> so, in short, yes, you are right :)
[19:41:08] <Denny_WMDE1> one reference should bind the URL and the accessdate together
[19:41:28] <Denny_WMDE1> it will not be ambiguous which accessdate belongs to which URL
[19:41:35] <Denny_WMDE1> that's how it is
[19:41:49] -*- Micru agrees
[19:41:49] <Denny_WMDE1> (besides the fact that we neither have URLs nor dates as a datatype yet...)
[19:42:03] <Denny_WMDE1> ok, i have one question
[19:42:10] <Denny_WMDE1> it's about book editions
[19:42:24] <Sven_Manguard> shoot :D
[19:42:30] <Micru> yes, i have my mind clear about that too, shoot :)
[19:42:31] <Denny_WMDE1> we have an item for The art of computer programming
[19:42:41] <Micru> (first edition)
[19:42:48] <Denny_WMDE1> and now i want to cite the third edition, page 234
[19:42:52] <Micru> new item
[19:43:01] <Denny_WMDE1> every edition that we cite one item?
[19:43:11] <Denny_WMDE1> i mean, don't they even have the same isbn and stuff?
[19:43:12] <Sven_Manguard> item as in Q item?
[19:43:19] <Sven_Manguard> no, that's wack
[19:43:21] <Micru> yes, the main item represents both the work and the first edition, the other editions need another item
[19:43:27] <legoktm> no
[19:43:30] <Denny_WMDE1> uhhhm....
[19:43:32] <legoktm> i dont think thats feasible
[19:43:40] <Sven_Manguard> agree with legoktm
[19:43:50] <legoktm> every wikipedia will just have 1 article per "book" there isnt a separate article per edition
[19:43:58] <Micru> legoktm, if you want all editions in one item, then we need a way to address one edition or the other
[19:43:58] <Denny_WMDE1> (in most cases)
[19:44:02] <legoktm> the item should list all the different ISBNs that it might have
[19:44:10] <legoktm> Micru: so we just have an "edition" property
[19:44:15] <Denny_WMDE1> yeah
[19:44:18] <Sven_Manguard> we should specify the number of editions as a statement on the book
[19:44:28] <Sven_Manguard> we should specify the number of editions as a statement on the item for the book
[19:44:37] <Denny_WMDE1> maybe
[19:44:42] <Micru> legoktm, how do you tell appart one edition from the other? old books have no isbn
[19:45:06] <legoktm> year of publication? # of pages?
[19:45:13] <Sven_Manguard> Micru: and really old books don't even identify their edition
[19:45:21] <legoktm> if you can't tell the editions apart, aren't they the same thing? :P
[19:45:33] <Micru> legoktm, not really :)
[19:45:40] <legoktm> :P
[19:45:59] <Micru> and sometimes items have translations which, in a way, are also editions
[19:46:19] <legoktm> hm, translations would be interesting
[19:46:34] -*- legoktm goes to propose a new property
[19:46:42] <Micru> yes, either a way of telling them appart in the same item is needed, or then separate items
[19:46:54] <Denny_WMDE1> could say stated in -> TAoCP, translation -> German, Edition -> "2", page -> "123"
[19:47:01] <Micru> legoktm, i think in the books task force is already suggested
[19:47:19] <Micru> what is TAoCP?
[19:47:41] <Denny_WMDE1> sorry, an The Art of Computer Programming
[19:47:46] <Denny_WMDE1> just used it as an example item
[19:47:57] <Micru> and well, if you can guarantee that the internal number for the edition won't change, then yes, it could be possible
[19:47:59] <Denny_WMDE1> because its a book where the editions are quite different
[19:48:34] <Denny_WMDE1> hmm. we could also be pragmatic
[19:48:42] <Denny_WMDE1> and say, in most cases, editions dont really matter
[19:48:51] <Denny_WMDE1> they can just be given in the reference, and thats it
[19:48:57] <Denny_WMDE1> but in a few cases
[19:49:12] <Denny_WMDE1> heck, it is really important to know these three different editions of the bible
[19:49:19] <Micru> exactly
[19:49:36] <Micru> we have been going through this over and over in the book task force discussions
[19:49:51] <Denny_WMDE1> so i guess we should just delegate the answer to the task force? :)
[19:49:56] <Micru> :D
[19:50:19] <Denny_WMDE1> ok, but it seems we are already much clearer in everything than we have been a few weeks ago
[19:50:19] <Micru> well, there are two sensible options, either to have edition IDs or to have separate items
[19:50:36] <Micru> yes, it is more clear
[19:50:37] <Denny_WMDE1> Micru: but it doesnt have to be one decision for all sources
[19:50:41] <Sven_Manguard> I am not in favor of separate items
[19:51:04] <Denny_WMDE1> I agree with Sven_Manguard and legoktm that in most cases we should not have separate items for editions
[19:51:05] <Micru> Sven_Manguard, I also prefer all in one, but it has to be done right to avoid problems
[19:51:21] -*- legoktm just proposed
[19:51:34] <Denny_WMDE1> ok
[19:51:41] <Denny_WMDE1> my goal was to catalyst the discussion a bit
[19:52:00] <Denny_WMDE1> i will post the log somewhere, i guess
[19:52:08] <Denny_WMDE1> but i would love to head off home now :)
[19:52:09] <legoktm> !logs
[19:52:20] <Denny_WMDE1> anyone has any more questions?
[19:52:20] <legoktm> sure :)
[19:52:25] <Denny_WMDE1> legoktm: what does !logs mean?
[19:52:26] <legoktm> thanks for hosting this
[19:52:35] <legoktm> i thought the bot would give a link to the log :P
[19:52:45] <Micru> ok, we can keep talking another day :
[19:52:51] <legoktm>
[19:53:46] <Denny_WMDE1> or on wiki :)
[19:53:48] <Denny_WMDE1> see you all!
[19:53:54] <Denny_WMDE1> it was a pleasure
[19:53:58] <Micru> bye! thanks for the time1