Wikidata talk:WikiProject Books/2019

a bit more complex items

Hello folks, I would like to start creating the book complex of Die Kunstdenkmäler der Rheinprovinz (Q1214163), which is a kind of monographic series, a monuments inventory which consists of multiple issues which are numbered hierarchically. de:Die Kunstdenkmäler der Rheinprovinz might give you an idea of the structure … We have

• Band 1-20 (volumes?)
  • every Band has issues called "Abteilung" (with its own title)
  • sometimes we have additional "issues" (?) called "Ergänzungsband" (called band, but on the level of issues)
    • see also Band 17, Abt. 2, Halbbd. 1, which means: volume 16, issue 2, split up in 2 sub-issues
  • not to mention an "Abt. 1/2", connected issues in one item

These items could and would be needed to set references for a lot of statements, so I'd like to to define them as precise as possible, in the best case to add just a page number. I tried to use only the top level title with numbers as reference, but this is not very satisfying (and I'm not sure it's understandable/traceable): Q15428477. I started one issue Die Kunstdenkmäler des Kreises Kempen (Q61456788) to give an example of an issue.

So, is there something like a best practise for this kind of … complex stuff? Thanks for your suggestions. --Elya (talk) 22:22, 3 February 2019 (UTC)

@Elya: You may want to look at what I have done with the Berg Encylopedia of World Dress and Fashion, published in 10 volumes in hardcover and online, with DOIs for each article.

• Berg Encyclopedia of World Dress and Fashion (Q4891400) (work}
• Berg Encyclopedia of World Dress and Fashion (Q55816150) (edition with "has part" for each volume)
• Berg Encyclopedia of World Dress and Fashion: Africa (Q55816186) (example volume)
• kanga (Q942219) (item with "described by source" and individual statement references to a specific article in a volume).

Not sure if this is a "best practice" but it works for me. - PKM (talk) 00:05, 4 February 2019 (UTC)

@PKM:, ah, this looks promising, thank you. This edition thing is still a bit strange for me, but the series ordinal (P1545) and subtitle (P1680) definitely will help. And the idea of just creating sub-editions on a second level – one could do that recursively with the same properties. A is part of B which is part of C, A has series ordinal (P1545) = 1 as part of B, B has series ordinal (P1545) = 1 as part of C etc. I'll give it a try. --Elya (talk) 05:59, 5 February 2019 (UTC)

@PKM:Well, this was interesting. I'm down to the (above not mentioned) reprint editions like Die Kunstdenkmäler des Kreises Euskirchen (Q61695140), which are the only editions with DNB edition ID (P1292) and ISBN-13 (P212). They are very likely to be the editions that will be referenced, if not the digitized versions. The last thing I did not solve yet is the fact that these reprints in some cases are combining different editions from the parent editions – (it's normal that you get a headache in this moment …). And there are a couple of digital/BOD reprints I left out for the moment.

Would be interesting to visualize these structure in some kind of tree view … Again, thanks for your help! --Elya (talk) 06:33, 12 February 2019 (UTC)

@Elya: "these reprints in some cases are combining different editions from the parent editions" What is the problem with that case ? Just consider that these reprints are another new edition. The only problem is if you want to keep the information that one edition is a mix of some other editions. But is this information really helpful ? I don't think so. Most of the items about books in WD are not properly defined and most of them lack the minimal necessary information, so I have some doubts that the relation of one edition with other ones is critical. Snipre (talk) 13:21, 12 February 2019 (UTC)

Pseudonym item: adding statement about possible identity

Hi, I asked this question on the Community Portal and was told that this is the right place to post, so I'm repeating myself (with modifications due to reading your past discussions). Namely, there is an author, Louis Sand (Q61550626) (Wikisource page); the Curran Index discovered thanks to a magazine's account books that the name is a pseudonym for one "Miss Lucy Smith." They suggest rather doubtfully that this may be Lucy Toulmin Smith (Q15439811). How should I enter Louis Sand in Wikidata? Currently I have nothing on that page except "instance of pseudonym." One possibility would be to add some sort of statement (what?) relaying the speculation about the real identity. Another possibility would be to upgrade the Louis Sand page to full author status; after all, this name is the identifier uniting a coherent and rather extensive body of work, and it can't be definitely synonimized with another entity at this point. In that case, are there statements I could add about 1. being a pseudonym and 2. speculations about identity? Levana Taylor (talk) 15:55, 6 February 2019 (UTC)

@Levana Taylor: said to be the same as (P460) could be used, with qualifier statement supported by (P3680) available to indicate supporters of the proposition, and statement disputed by (P1310) available to indicate those that reject it. Jheald (talk) 18:09, 1 March 2019 (UTC)

How to translate participants secction?

If I use cs language, I can see in Participants section this "Participants (en) – (Toto prosím přeložte do jazyka čeština.)", which means, Translate this to Czech language please. But if I click on translate of the main page, I dont have this secction there. So where or how can I translate it?--Juandev (talk) 10:02, 27 February 2019 (UTC)

Size

Do we log size? E.g. wheater the size of the book is A4, A5, or 12 x 23 cm?--Juandev (talk) 10:15, 27 February 2019 (UTC)

@Juandev: If you want to give the measurements you can just use height (P2048), width (P2049) and thickness (P2610).

I had a similar question though: How is it possible to give an item for the format? This would be useful with the values folio (Q772267), quarto (Q2122442), octavo (Q1307353) etc. (instance of (P31)folio book (Q13636757) would be possible, too, but isn't a very good solution I think.) Thanks for any help! --Marsupium (talk) 14:57, 27 February 2019 (UTC)

distribution format (P437) is available to specify the distribution format -- eg quarto, octavo, duodecimo -- though a few entries might need to be added to the list of allowable values. Jheald (talk) 18:05, 1 March 2019 (UTC)

Short story reprints

This item Win a Nobel prize! (Q48371781) is for a short story published in an issue of Nature in 2000, with statements for volume, page number, issue, and date of publication. But how should it be recorded that the same story is published in The Collected Stories of Vernor Vinge (Q7726674)? Ghouston (talk) 10:46, 23 March 2019 (UTC)

You would create a new data item for the edition that was published in The Collected Stories of Vernor Vinge (Q7726674), and both the new data item and older data item wouldbe identified as an instance of (P31) version, edition or translation (Q3331189) of a data item for the story as a "work". --EncycloPetey (talk) 14:49, 23 March 2019 (UTC)

OK, thanks. Ghouston (talk) 23:35, 23 March 2019 (UTC)

Wikidata property to identify books

SWB editions (P1044) will be integrated into the new catalog K10plus. Do we need a new property or should the old numbers (SWB-ID) be replaced by the new numbers (K10plusPPN)? Example:

• Die natürlichen Pflanzenfamilien, 2nd edition (Q5805827), SWB editions (P1044): 000619612
• Die natürlichen Pflanzenfamilien, 2nd edition (Q5805827), K10plusPPN: 1070619612

--Kolja21 (talk) 11:22, 24 March 2019 (UTC)

See: Wikidata:Property proposal/K10plus editions --Kolja21 (talk) 23:07, 26 March 2019 (UTC)

Number of prints

Do we want to have property for the number of prints of a specific book? I havent found it here.--Juandev (talk) 10:25, 27 February 2019 (UTC)

@Juandev: total produced (P1092) is available. Jheald (talk) 18:13, 1 March 2019 (UTC)

Fantastic!--Juandev (talk) 16:13, 28 March 2019 (UTC)

General vs. specific instance of a published work

I have a newbie question... I was looking through the list of properties related to books, and it seems like they're mostly aimed at identifying specific editions of books. If I entered, say, the first edition of a novel as a "work", is there a way I could also add a page for the non-edition-specific novel-per-se, which would contain info about its setting and so forth that would be shared by all editions? Levana Taylor (talk) 23:17, 28 February 2019 (UTC)

@Levana Taylor: Sorry but I don't understand your question: the wikidata model implies you create 2 items, one for the work and one for the edition. So you can't enter the first edition of a novel as a "work", when you want to add a book to wikidata because no item exists about it, you create first the work item with the data about the work (common data to all editions), then you create second item for the specific edition you want to describe. If the work item already exists because another edition was already added, then you just create the edition item for the edition you want to describe. See as example CRC Handbook of Chemistry and Physics (Q904273) as work item and CRC Handbook of Chemistry and Physics (92nd edition) (Q11927173) as edition item. Or A Fire Upon the Deep (1st edition) (Q74335) as edition item and A Fire Upon the Deep (Q58221164) for the work item. Snipre (talk) 16:55, 1 March 2019 (UTC)

More bad genre edits

User:Marcok has incorrectly added genre (P136) short story collection (Q1279564) to many, many data items. short story collection (Q1279564) is not a genre, and so is not a valid value. Further, many of these works are themselves a single short story (Q49084), and not a collection at all. (see for example Q60854160.)

I left a message on Marcok's talk page, but he seems to have been offline since mass adding the values.

Is there someone with a bot who could edit all these items to remove short story collection (Q1279564) as a "genre"? --EncycloPetey (talk) 00:22, 20 March 2019 (UTC)

Alternatively, why don't we just change the label of property genre (P136) to be "genre or form" ? Jheald (talk) 22:59, 27 March 2019 (UTC)

Because (a) that affects multiple projects, and (b) the "form" is handled as a value for instance of (P31), assigned by FaBiO. Conflating the two concepts even further would not improve the quality of the data. --EncycloPetey (talk) 23:24, 27 March 2019 (UTC)

This query generates instructions to remove all such statements. The results just need to be pasted into Quickstatements:

SELECT (CONCAT("-",SUBSTR(STR(?item),32)) AS ?s) ("P136" AS ?p) ("Q1279564" AS ?o) WHERE {
?item wdt:P136 wd:Q1279564
}

Try it!

MartinPoulter (talk) 14:19, 4 April 2019 (UTC)

Manuscripts

Hi, is there an agreed way of stating the authors of individual works contained within a manuscript - or should these simply be created as separate items and linked with (Part of)? Thanks! Jason.nlw (talk) 09:49, 2 April 2019 (UTC)

@Jason.nlw: Without being an expert of this WikiProject I'd guess they should best be created as separate items and get linked with exemplar of (P1574), that's at least how I've modeled it so far. --Marsupium (talk) 23:13, 2 April 2019 (UTC)

Thanks Marsupium, that's kind of what i was thinking, it just gets a bit messy when you have volumes containing dozens, if not hundreds of poems by 10 or more authors! Jason.nlw (talk) 08:37, 3 April 2019 (UTC)

@Jason.nlw: I don't understand your question: "stating the authors of individual works contained within a manuscript", just use author (P50). Snipre (talk) 06:15, 3 April 2019 (UTC)

@Snipre: Sorry, i should have been more clear. For example if a manuscript contains the work of Geoffrey Chaucer, but is in the hand of (written/copied by) Adam pinkhurst, How do i say that Chaucer is the author of the original work? The guidelines in this project suggest Adam Pinkhurst would be the author (P50). I am preparing a collection of manuscripts for upload, and the data for many gives names of scribes, and the names of the authors of the works the manuscript contains. Many contain dozens of poems by different authors. I hope that makes more sense. Thanks! Jason.nlw (talk) 08:23, 3 April 2019 (UTC)

@Jason.nlw: Ok, so if I understand well, we should have

• one item about the work with p50 = Geoffrey Chaucer
• one item about the manuscript written by Geoffrey Chaucer with p50 = Geoffrey Chaucer
• one item about the manuscript written copied by Adam pinkhurst with p50 = Geoffrey Chaucer

So the conclusion is we need a new property to mention who is the copist of a manuscript like we have a property for the translators (see translator (P655)). Snipre (talk) 09:25, 3 April 2019 (UTC)

Hi Snipre, yes a new property would make sense i think. Most Medieval manuscripts are copies of original works and often no copy in the hand of the author exists. I Kind of think that p50 should be used like it is with books - to state the author of the work or works that book contains, and copyist should be used in the same way that Publisher is used on printed books, to state who is responsible for physically producing the work. Jason.nlw (talk) 08:47, 4 April 2019 (UTC)

@Jason.nlw, Snipre: See Wikidata:Property proposal/calligrapher where I've voted to use creator (P170) with qualifier object has role (P3831). --Marsupium (talk) 11:25, 4 April 2019 (UTC)

Hi Marsupium - thanks for pointing me to this discussion. user:sic19 has also suggested using creator to state the name of the scribe/copyist/calligrapher. Would it then be appropriate to use P50 to state the authors of the literary works contained within the Manuscript? Jason.nlw (talk) 11:56, 4 April 2019 (UTC)

How to handle multiple publications of the same scholarly article?

Some scholarly article are published in different journals or are republished by the same journal. What to do in such cases? Treat the scholarly article such a book, treating each publication as an edition and creating an item for the article itself?--Malore (talk) 21:31, 2 April 2019 (UTC)

Yes we have to create a new item for the new publication, similar to different editions of a book. What is missing perhaps is a property to link the different publications of the same article. Snipre (talk) 06:13, 3 April 2019 (UTC)

@Malore: it usually should not exist. Can you please give an example? --Infovarius (talk) 12:28, 4 April 2019 (UTC)

@Infovarius:First draft of a report on the EDVAC (Q56637246) is a 1993 republication of a 1945 article. In this specific case, the article is republished because it is a particular important one, but articles can be published by different journals because the author tries to reach a greter audience. There is also a wikipedia article about it.--Malore (talk) 22:34, 4 April 2019 (UTC)

Pages property

Can I use page(s) (P304) to indicate on which pages the article in the magazine could be located? I am sorry, but I am not sure/the property description is not clear to me.--Juandev (talk) 05:53, 9 April 2019 (UTC)

@Juandev: Yes, that's an appropriate use per latest discussion! --Marsupium (talk) 14:45, 9 April 2019 (UTC)

Good example of a work with multiple editions?

Might someone add an example of a work with multiple editions?

The effects of nuclear weapons

I ask, because I'm having problems with the following:

* The effects of nuclear weapons (Q63072754): 1957 book

* The effects of nuclear weapons (Q63079864): written work produced in 4 editions: 1950, 1957, 1962, 1977 (with at least the 1950 edition produced in sub-editions)

What needs to happen to fix these two, so each points appropriately to the other, and neither has "potential issues"?

A Wikipedia article (w:Samuel Glasstone) mentions all three editions. Conveniently, all three editions are US government documents, which means they are in the public domain. Moreover, scanned images of all three are available. However, that Wikipedia article does not link to any of these electronic copies. I can create such links via Wikidata, but I'd like to understand first the preferred way of doing this.

@DavidMCEddy: See CRC Handbook of Chemistry and Physics (Q904273) (work) and CRC Handbook of Chemistry and Physics (95th edition) (Q20887890) (edition) or CRC Handbook of Chemistry and Physics (97th edition) (Q42107747) (edition). Snipre (talk) 16:23, 10 April 2019 (UTC)

@DavidMCEddy: Here's another example: Patronage in the Renaissance (Q63041706). - PKM (talk) 23:34, 11 April 2019 (UTC)

Thanks. I've updated Q63079864 consistent with what I saw in those examples.

I also found a 1950 edition with a slightly different title that was actually published in two editions in 1950. I found a link to the second one and added that as "edition number 0.5". The software didn't like "edition number 0.5", but I'm not sure the best way to handle that, and I don't want to spend more time on this at the present time. I mention it in case someone else would like to suggest a preferred way to handle that. DavidMCEddy (talk) 05:22, 14 April 2019 (UTC)

Brockhaus Enzyklopädie

It might help if someone looked at Brockhaus Enzyklopädie (Q237227) and added a section to this article describing the 5 flags I see there now (2019-04-10) saying, “This statement has some potential issues". I'd like to see an explanation of how to fix each. Then, of course, it would be good to actually fix those issues with Q237227. The documentation here could help others fix similar problems with other Wikidata items.

Thanks, DavidMCEddy (talk) 16:10, 10 April 2019 (UTC)

@DavidMCEddy:, did you mouse over the flag to see the message? For example: Potential issue: value requires statement constraint. Brockhaus Enzyklopädie (14 ed.) should have a statement publication date.needs to be fixed on the linked item, not on Q237227. In this case, add a publication date to the linked edition to clear the error. All three editions need the same fix.

For the error on the publisher Leupold (Q968343), the message is telling you someone has chosen this item as the publisher, but it does not have a statement <instance of> publisher. The fix is more complex here, because in fact Leupold (Q968343) is an item for a Wikipedia disambiguation page, not for an item about a publisher. The current link should be removed, as it is incorrect. Wikidata does not appear to have an item about a publisher called "Leupold". You may create one if you have enough information to do so, or just skip the publisher field for now. - PKM (talk) 23:31, 11 April 2019 (UTC)

Publication date for digitized book

For a book digitized by the Internet Archive (like The Early Christian Monuments of Scotland (Q63131046)). should the <publication date> be the date the book was published or the date the scan was published? - PKM (talk) 02:11, 14 April 2019 (UTC)

Use the date that the original of the scan was published. We don't distinguish between an IA scan, and Hathi Trust scan, a Google scan, or a Wikisource copy built from such a scan when they are all the same edition. Also, we don't call the data item a "digitized" copy, because a physical copy of the same edition in a library can apply to the same data item. --EncycloPetey (talk) 03:02, 14 April 2019 (UTC)

Thanks, that works for me - although OCLC seems to have different control numbers for the scan and the print book. - PKM (talk) 06:42, 14 April 2019 (UTC)

@PKM: The British Library do this too, with different British Library system number (P5199) identifiers for the print book and the scan. I try to distinguish the statements with qualifiers subject has role (P2868) = physical object (Q223557) and subject has role (P2868) = digital representation (Q42396623). Jheald (talk) 17:53, 14 April 2019 (UTC)

@Jheald: That’s an interesting approach. I’d rather have a property like “applies to version” or “applies to manifestation” but I’m not sure what would be the best wording. For now, I have followed your lead wth this item. - PKM (talk) 20:17, 14 April 2019 (UTC)

@PKM: OCLC are inconsistent in much of what they do. You will find that their "Edition" (M) numbers sometimes apply to a particular scan, and at other times apply to all copies of a particular edition. --EncycloPetey (talk) 20:33, 14 April 2019 (UTC)

Invitation to join Linked Data for Libraries Wikidata Affinity Group First Meeting 2019-04-23

I'm a new Wikimedian in Residence as part of the Linked Data for Production project. One of the goals of the project is understanding how libraries can contribute to and leverage Wikidata as a platform for publishing, linking, and enriching library linked data. A number of institutions that are part of the grant are working on projects involving Wikidata and we decided to start an interest group with biweekly meetings to discuss various aspects of Wikidata in support of the projects. Possible topics include Wikidata best practices, documentation, communication channels, policies, and tools.

At each meeting, myself or a guest will present some relevant material related to the topic and we’ll discuss any issues members have encountered as well as helpful resources. At the first meeting on April 23rd, we’ll talk about the purpose and goals of the group as well as the Wikidata related projects that are part of the grant.

The Affinity Group is open to anyone interested. The call details and communication channels are below.

First call details:

• April 23, 2019 9am PST / 12noon EST / 5pm GMT / 6pm CET

• Agenda: https://docs.google.com/document/d/1BuszEQQxlOY14hK60Fl7n8Huvh6jEWXre0-wSvpyq84/edit?usp=sharing

• Join: https://stanford.zoom.us/j/204437188

Communication:

• ld4-wikidata Google group: https://groups.google.com/d/forum/ld4-wikidata

• Notes in public LD4 Wikidata folder: https://drive.google.com/drive/folders/1JwTulCABs0TkGQDVSnYbIYEb7bC-j4-n

• Website: https://wiki.duraspace.org/display/LD4P2/Wikidata+Affinity+Group

Chicagohil (talk) 21:43, 22 April 2019 (UTC)

@Chicagohil: What we need from professionals is a classification system for books, if possible with a list of advantages or disadvantages of the system. The current system presented on the project page is not a global concept but the aggregation of solutions to different problems. What is missing is the relations between the different levels of the system: most of the time WD contributors focus on one item (work, edition, exemplar, manuscript,...) but never try to organize the data distribution between the different levels to avoid duplication of data or dicrepancies. In my opinion we should not have several tables, one for each level, but an unique table providing the properties to use to describe one level and the ay the data can be retrieved from the different levels. Snipre (talk) 23:36, 22 April 2019 (UTC)

Key words

Can I use main subject (P921) for key words? Key words are few words, which characterize the content of a book or an article (e.g. Jaromír Jágr - interview - hockey - NHL). Juandev (talk) 07:14, 25 April 2019 (UTC)

Imho yes. --Kolja21 (talk) 15:33, 25 April 2019 (UTC)

Scored book reviews

Hi. I'm the Wikipedia Visiting Scholar at Northeastern University in the Women Writers Project. I'm interested in modelling pre-XX-c written works by women (novels, poems, essays, etc.), to include the reviews these works received, and whether the reviews were Positive, Neutral, or Negative. I appreciate any feedback on how to go about this. Thanks. --Rosiestep (talk) 20:42, 24 April 2019 (UTC)

Hi Rosie, I don't think there is a property for book reviews. Look at the review Der Kaiser ging ... (Q19157726) by Carl von Ossietzky. For the link to the book he reviews I've used main subject (P921). --Kolja21 (talk) 23:39, 24 April 2019 (UTC)

We have described by source (P1343), which can be used to identify the periodical or newspaper that included a review, and then qualify that with statement is subject of (P805) to add a data item containing the full bibliographic information for the particular review. --EncycloPetey (talk) 15:38, 25 April 2019 (UTC)

Sounds interesting. Can you give an example? What irritates me: statement is subject of (P805) is marked as "qualifier only". --Kolja21 (talk) 16:08, 25 April 2019 (UTC)

I've mostly seen this applied to people, such as Sophocles (Q7235). Look for the described by source (P1343) section. What happens is the primary value identifies the periodical / work in which the description (review) was printed, and the statement is subject of (P805) is used as a qualifier. The statement is subject of (P805) link then points to a data item for the article / review where all the publication details would be placed so that it can be cited.

So, if you had a review published in The Times (Q50008), then described by source (P1343) would have a value of The Times (Q50008). You would create a separate data item for the specific review article, and add it as a qualifier value using statement is subject of (P805). --EncycloPetey (talk) 20:20, 25 April 2019 (UTC)

Ok. Let's do it the same way as in the case of biographical articles. I've created an example:

• Der Kaiser ging, die Generäle blieben (Q63285462) novel → Der Kaiser ging ... (Q19157726) review

--Kolja21 (talk) 22:09, 25 April 2019 (UTC)

Thanks for the quick response. This would work for how to cite the review. Part B of my conundrum is that I want to "score" the review, e.g. that it was Positive, Neutral, or Negative. Could I do that (how to deal with the issue of bias?), and if it is possible, how? Adding Maxlath to the conversation per suggestion by Will (Wiki Ed) who is helping me with the brainstorming. --Rosiestep (talk) 00:22, 26 April 2019 (UTC)

hi! for the link between a review item and a work, in absence of a dedicated "review of" property, I guess main subject (P921) is the way to go: it's already in use here and there for reviews. As for the score, we might lack an adapted property: we already got review score (P444), but it seems that it is used on the reviewed works to indicate a score and then add the review source as a qualifier with review score by (P447). -- Maxlath (talk) 08:29, 26 April 2019 (UTC)

@Kolja21: You can't use instance of (P31) book (Q571), because "book" / "Buch" is a nonspecific label. It could refer to a work, a specific physical copy, or anything in between. Use literary work (Q7725634) or one of its subclasses. --EncycloPetey (talk) 01:00, 26 April 2019 (UTC)

@Maxlath:, @Rosiestep: Is there a way to capture the sentiment of a review if it's not numerically-based? A review inherently contains a judgement on a work, but without a numerical evaluation system, I can't think of a non-subjective way to evaluate the conclusion of a review. This project sounds so compelling, but I don't want it to fall into original research. I would be in favor of creating a "review of" property! Will (Wiki Ed) (talk) 17:56, 29 April 2019 (UTC)

+1. I would support a "review of" property. --Kolja21 (talk) 18:01, 29 April 2019 (UTC)

I’d support this too. There are more reviews wth the subject book as “main subject” than the query above captures - some are P31=“scholarly article” with genre=book review. - PKM (talk) 01:47, 30 April 2019 (UTC)

Hi everyone - @Rosiestep: just proposed the "review of" property. Feel free to make a comment here on the property proposal page. Will (Wiki Ed) (talk) 22:04, 2 May 2019 (UTC)

Thanks, everyone, for your comments here and if you choose to make them on the proposed "review of" property. If all of this goes forward, it'll be a pretty big project for me to undertake. BTW, does anyone know if/how the Women Writers in Review API can be transformed into Wikidata? --Rosiestep (talk) 22:58, 2 May 2019 (UTC)

Rubric/column

Do we have any property, which would represent this? rubric (Q1335351) For example in the newspapers, some articles are under a special name, called sometimes rubric.--Juandev (talk) 19:21, 25 April 2019 (UTC)

Not sure what you're seeking to represent. I've never seen a newspaper that printed a section in red ink. Normally, I see this only in printed copies of the New Testament. --EncycloPetey (talk) 20:16, 25 April 2019 (UTC)

So, lets not talk abour the rubric. Lets talk about something, I dont know, how it is officially called in English. And lets go into the example: the magazine is called Forbes, the XY is called Tech - the article is called "New Tesla is coming". I can set published in (P1433) = Forbes, but how can I indicate "New Tesla is coming" article is a part of XY Tech?--Juandev (talk) 11:10, 26 April 2019 (UTC)

We might create a Wikidata item for “XY Tech” with <instance of>column (Q267628), and then use <published in> Forbes, <part of> “XY Tech”. I’m not sure about <part of> for an article in a recurring column. What do others think? Do we need a new property for this? That might be the cleanest solution. - PKM (talk) 04:30, 27 April 2019 (UTC)

The problem is, we in Czech use for this the term rubric, but there is no Czech description for rubric (Q1335351) and if I look to w:en:rubric it does not use that sence either, but Czech w:cs:rubrika does: and it says rubric is a section in the magazine or newspaper. So I wonder, weather the closest way of the Czech "rubrika" is "section" in English.

So looking for properties which include the term section, I came to section, verse, paragraph, or clause (P958), which may stay for that.

E.g. "New Tesla is coming"

• <instance of> news article (Q5707594)
• <published in> Forbes
• <section, verse, paragraph, or clause (P958)> Tech

Could it work like that? Juandev (talk) 08:48, 27 April 2019 (UTC)

Have you asked at Wikidata:WikiProject Periodicals? This seems like a question better suited to the works they are working with than it does with books. --EncycloPetey (talk) 17:55, 27 April 2019 (UTC)

Oh, thx, I havent been aware of this project. --Juandev (talk) 20:29, 27 April 2019 (UTC)

How does Has Part interact with the work/edition distinction?

The project page mentions has part(s) (P527), but completely fails to clarify whether this property is supposed to be used on the work item or the edition item. How do I connect the various preprinted monographs of Histoire des Plantes (Q5898798) to it (e.g. Histoire des plantes. Monographie des Renonculacées (Q51488450), which is the first chapter in volume 1 of that work)?

Am I going to have to make a work/version separation for each of these preprint (which, I would like to emphasize, do not otherwise have other versions independent of the greater work)? I would honeslty find that preporsterous and give up ever continuing my work with Taxonomic literature : a selective guide to botanical publications and collections with dates, commentaries and types (Q56649865)-indexed material. Circeus (talk) 06:44, 29 April 2019 (UTC)

Modeling the two editions of Taxonomic literature (Q56683530) here is an interesting case. But I don't think this enhancement for TL2 is correct. --Succu (talk) 19:00, 29 April 2019 (UTC)

I was actually referring to my going through the works in it to add Taxonomic Literature 2 number (P5878) and general enhancement and fixes. Even if one chooses to treat the TL-2 supplement as a separate work (I personally firmly believe both options are defensible, but obviously leans toward treating them as not separate), modelling them is nowhere as complex as Histoire des Plantes (Q5898798), where the work has these monographs (prepublished, but reproducing the larger work's pagination) and a volumation (Not to mention an english translation—The natural history of plants (Q51508335)—whose volumation is entirely different!). Circeus (talk) 19:18, 29 April 2019 (UTC)

BHL has only a subset of the 13 volumns: A temporary solution… --Succu (talk)

Beg your pardon...

Mind you, that doesn't answer my question about how I'm supposed to properly interconnect the ca. 50 monograph preprints, all of which have items on Wikidata (primarily because they are also found on BHL via Biblioteca Digital Real Jardin Botanico). Circeus (talk) 20:35, 29 April 2019 (UTC)

Am I'm right you want to model this TL2 entry? --Succu (talk) 21:03, 29 April 2019 (UTC)

What's the relationship between Histoire des plantes (Q51398585) and Histoire des Plantes (Q5898798)? --Succu (talk) 21:32, 29 April 2019 (UTC)

Homonymous book by Louis Figuier (TL-2 number 34.454) with no connection aside from sharing the same title and illustrator. Circeus (talk) 06:48, 30 April 2019 (UTC)

I used “has part” on the edition, not the work, with Berg Encyclopedia of World Dress and Fashion (Q55816150). - PKM (talk) 02:00, 30 April 2019 (UTC)

More broadly: should multivolume work always have volumes as separate part?

It seems that the main reason we see multivolume work with separate volume items is 99% of the time that these volumes were indexed separately in BHL. And by this I mean most such instances are bot-imports for items that were later combined under a single BHL bibliography id or are separate solely because they are external links (e.g. the various volumes of Dictionnaire de Botanique (Q5805676) which BHL does not host would be under a single BHL bibbliography ID if it did). A related side effects is duplicates caused by the Biblioteca Real that have otherwise no business existing such as Astragalogia, nempe astragali, biserrulae et oxytropidis :nec non phacae, colutae et lessertiae historia iconibus illustrata (Q51444464) and Astragalogia (Q51444463), indubitably two copies of the same edition of the same book, but merging them would cause issues with values of BHL bibliography ID (P4327).

In general, it seems to me that volumation is a bibliographic detail of the relevant edition that pretty much only becomes relevant when the item is used as a reference. In practice, there is no difference between using a volume that is a separate item or a version/edition/translation item + volume (P478). Indeed, we do that for periodicals, and why wouldn't we? We wouldn't consider it logical to have separate items for volumes of a periodical. When we do have them, it is for the same reason we do volumes of a multivolume work: they are bot-generated.

Really, the only reason we don't do this is that there no string-data version of applies to part (P518). With an "applies to volume" property (heck, series ordinal (P1545) could handle it if it didn't clash occasionally or need to handle named instead of numbered volumes), the only property that IMO genuinely causes issue for Taxonomic literature : a selective guide to botanical publications and collections with dates, commentaries and types (Q56649865) (i.e. ISBN-10 (P957)) could be neatly treated under a single item and eliminate the need for volume items entirely (this could cover all other properties that vary across the work just the same). Circeus (talk) 12:41, 30 April 2019 (UTC)

@Circeus: If you look at Wikidata:WikiProject_BHL/Statistics:Titles#Titles_--_IDs, you can see we have 58,938 Internet Archive IDs for only 53,989 works. In most cases that already exactly corresponds to what you're describing, a single item for a multivolume work edition, with multiple Internet Archive IDs, a different one for each volume. So in very many cases, including BHL works, we already do pull all the volumes together under a single item, despite there being different links to sites to read them.

That's also what I am doing for books from the British Library: See Wikidata:WikiProject_BL19C/Statistics#Titles_--_statements and its different item and statement counts for full work available at URL (P953).

I don't see a need for a new "applies to volume" property -- The existing volume (P478) can and indeed already is fulfilling this role when used as a qualifier. (And is string-valued, so can deal with non-numerical parts).

In fact here is a query counting how many times volume (P478) is being used as a qualifier, and on what properties, showing heavy use on Internet Archive ID (P724), full work available at URL (P953), publication date (P577), document file on Wikimedia Commons (P996), Google Books ID (P675) to distinguish different values for different volumes. Jheald (talk) 14:23, 30 April 2019 (UTC)

Part of my thought was predicated on the presumption that volume (P478) asked for a numeric value. I'm glad to be wrong in that. I will have to start merging items, though in some cases it gets really hairy (I've stumbled across one particular multivolume work with the same volume having 3 different Internet Archive scan, and even if BHL merges the 4 bibliographies, this duplicate scans remain a likely constraint violation). Circeus (talk) 14:42, 30 April 2019 (UTC)

@Circeus: I don't think volume items have to be merged into the item for the edition, if it is more convenient to treat them separately. In some cases, also, it may be more appropriate to treat the overall entity as a series of creative works (Q7725310) -- often the line can be a bit grey. The important thing, I think, is that in the converse direction there is definitely no requirement to split a multi-volume book into separate items for each volume. Jheald (talk) 14:52, 30 April 2019 (UTC)

I prefer to treat volumes as separate items when each volume has its own subtitle and when each volume has a different editor or author. - PKM (talk) 01:59, 8 May 2019 (UTC)

Also for many dictionaries I've dealt with so far the volumes have different publication date (P577). I prefer separate items as well, but don't always create them tbh because it's more work. --Marsupium (talk) 00:19, 11 May 2019 (UTC)

Connecting Wikisource related pages on Commons to Wikidata

Cross posting with Wikidata talk:Wikisource:

Commons has 779 thousand pages calling Book infobox template, mostly used by files used by Wikisource projects. We have over 2 thousand templates in c:Category:Book templates, which are mostly used for books that store each page in a separate file and great many PDF and DjVu files storing multiple pages. Very few of those books, are connected to Wikidata in any way: few sitelinks from Wikidata book items to Commons categories and very few Book templates with "wikidata" parameters linking them to Wikidata. Lately, I was working on rewriting Book template to use Lua code c:Module:Artwork, already used by Artwork and Photograph templates. The new code will be able to utilize data from Wikidata if "wikidata" parameter is provided to the template. I am not done testing it yet, but the new version can be accessed through Book/sandbox if anybody wants to try it. Also it would be great if there was some way to add "wikidata" parameters to Book templates and individual book files, as most wikisource entries are already connected. --Jarekt (talk) 20:28, 6 May 2019 (UTC)

@Jarekt: Naming your lua box Book is perhaps not the best idea: since the beginning of WD we are struggling with that term which mixes 3 concepts: work, edition and exemplar. I assume that in more cases, these boxes refer to editions and some to particular exemplars. It can be a good idea to separate these 2 cases using different boxes. Snipre (talk) 21:26, 6 May 2019 (UTC)

Snipre, I am not creating any new names. Book was created over a decade ago, and I am updating it so if connected to wikidata item it can use metadata stored there. If it is not connected (great majority at this point) than the template will work the same way as the original. --Jarekt (talk) 02:19, 7 May 2019 (UTC)

The problem is that the template "Book" might be used on Commons in a category for a WORK, or in a category for an EDITION, or in a category for an EXEMPLAR. Sometimes the category on Commons mixes two or three of these in a single category. Until Commons disentangles these three concepts from "Book", there is little that Wikidata can do. --EncycloPetey (talk) 00:17, 8 May 2019 (UTC)

I would not expect Commons users to understand the distinctions wikidata makes between works, editions, and exemplars.

@Jarekt: this looks like a rich and overlooked source. I’ll take a look at the sandbox. - PKM (talk) 01:37, 8 May 2019 (UTC)

Book is used to describe physical scans and they all have (or should have) edition numbers etc., so majority should be linked to EDITION items. Some may be unique manuscripts, and those should be linked to items for tat unique book (EXEMPLAR?). Categories for books on Commons are often underdeveloped, bordering on nonexisting, so I would not be using then to guide you, and they also can be easily changed. --Jarekt (talk) 02:23, 8 May 2019 (UTC)

@Jarekt: I understood your point, but even with an unique infobox, we can generate different looks depending on the information entered in the infobox. This could be a way to distinguish the infobox in function of the topic but using the same template. Snipre (talk) 09:35, 8 May 2019 (UTC)

The Book/sandbox already displays instance of (P31) property. I can highlight differences between work, edition and exemplar type items, or just group them to separate categories so it is easier to inspect them. I am doing something similar for artworks as I try to detect cases when people linked artworks to items for authors or subjects. --Jarekt (talk) 12:39, 8 May 2019 (UTC)

+1 for getting them into dedicated maintenance categories. Here, we have the same problem as for art prints and a similar problem to that with groups or series of artworks. Institution/collection data e.g. should in most cases only be in exemplar items.

It could be useful to put the most special or lowest level Wikidata item available as the |wikidata= parameter value on Commons, fetch information not in that item from higher items which can be an edition and a work item. If Commons holds information on a "book" in a lower level than the Wikidata item for the lowest level new items for lower levels should be created if we consider them to be notable enough. Otherwise that information should live on Commons. --Marsupium (talk) 00:46, 11 May 2019 (UTC)

New IDs mostly about Italian authors

Hi y'all. We have discussed with User:Epìdosis about some new IDs for book authors availabe on the publisher sites, mostly Italian. He made this accurate list:

• Bompiani (https://www.bompiani.it/autori)
• Einaudi (https://www.einaudi.it/autori/A/)
• Feltrinelli (http://feltrinellieditore.it/autori/letter/A/)
• Garzanti (https://www.garzanti.it/autori/)
• Giunti (https://www.giunti.it/autori?query=*)
• Marsilio (http://www.marsilioeditori.it/lista-autori/autori)
• Mondadori (https://www.librimondadori.it/autori/)
• Rizzoli (https://www.rizzolilibri.it/autori/)
• [Pisa University Press] (https://www.pisauniversitypress.it/elenco-autori.html)

I prefer to ask you first and not the Italy project in this case. What do you think? Also, is anyone interested in some semi-automated import? That's not really my cup of tea, so far I mostly help with manual updates and uploads during literacy events.--Alexmar983 (talk) 08:59, 9 May 2019 (UTC)

Just a little addition: one already exists, Adelphi author ID (P5859), and I have completely matched it. --Epìdosis 09:02, 9 May 2019 (UTC)

Good.--Alexmar983 (talk) 09:54, 9 May 2019 (UTC)

FYI if there is no opposition unless I have some doubts while scrolling their website, I will propose them one by one in alphabethical order.--Alexmar983 (talk) 09:55, 9 May 2019 (UTC)

@Alexmar983: What is the interest of adding so many new ids ? Can't we cover these authors with more general IDs ? My problem with your proposition is the fact that authors are not strongly connected to publishers. Publisher information are more related to editions and we can in any case establish the link between publishers and author through the editions items. I am not strongly opposed but I find these new properties useless. Snipre (talk) 14:44, 9 May 2019 (UTC)

I concur with Snipre: What is to be gained from these IDs? Do they add any value to Wikidata? --EncycloPetey (talk) 14:56, 9 May 2019 (UTC)

All these IDs contain (with few exceptions) some biographical information about the author (date of birth, but tipically also something else, e.g. the academic position or the alma mater, which are usually absent in more general IDs) and can also be useful for notability. --Epìdosis 17:16, 9 May 2019 (UTC)

(conflicted)They add the same value of any other ID. Acutally even more because they contain small biogrpahies so they can be used as effective sources for specific information. These are the main editors in italy, they are very accurate, reliable and they cover very well almost all the authors when combined. I am not understanding the part of the connection to publishers, but if there is a guideline that states a specific threshold for quality please let me know. So far, they look to me IDs like the others.--Alexmar983 (talk) 17:19, 9 May 2019 (UTC)

Not every ID is of equal value. Library databases are usually NPOV, presenting facts. The "biographies" that I've looked at from these Italian publishers look to be mostly promotional, and very POV. The purpose of these biographies is not to present facts, but to sell books. --EncycloPetey (talk) 17:55, 9 May 2019 (UTC)

They are not promotional. We are both expert users and we checked before proposing them, we did not notice any POV (not even very POV). Show an example please. Also, please show a guidiline according to which I can calibrate the different values of these IDs. As far as i can see they are reliable descriptions produced by established publishers, and they can be use as source... even in itwikipedia, one of the most strict regarding notability and quality source, they are considered reliable.--Alexmar983 (talk) 18:07, 9 May 2019 (UTC)

@Alexmar983, Epìdosis: The quality of the data is very different:

one example: no information

one example: no information

If I recognize that some information can be helpful, it is far away from more general database and perhaps less stable as commercial sites. Again I am not strongly oppose to the creation of new properties but if you really want to get some support better indicate the advantages of the related databases. For me the data found in these databases is too variable in term of quality (no systematic, raw text,...).

But finally you don't answer my first question: aren't all these authors already present in large databases which are authority in the field of litterature ? You can start the property proposals but don't be surprise if the support is low. Snipre (talk) 20:57, 9 May 2019 (UTC)

Snipre actually I reply at a certain point asking to show me a guideline that states how to calibrate the different values of IDs here. In my mind, this includes also to show me how guidline defines how superposition is a important aspect for an ID and how you would define it. Having a superposition with previous databases is not a problem for a new database, considering also there is no way in many cases to calculate it before importing it nd they often evolve (but there are tools we use to evaluate after the import and is sometimes even 90%-95%, and I don't recall any suggestion for removal in those cases). it's "emotional" more than factual. I could understand for very small sets that it's not worth the effort if they are mostly included in a big one, but these are quite big. How can you evaluate as critical something that you cannot objectively estimate? The point in my experience from a practical point of view is not if two databases are superposing (good database always superimpose on the long term, that's in a way a proof of their quality because the converge and expand) but if they are independent. Everything on scopus is basically on google scholar, to cite one example. So I don't expect the support to be low because of such aspect, many other users will understand this is not defined and could be applied to everything. If not, this is not a problem of the databases but the fact that there is a lack of a general perspective. That's why I use the verb "calibrate" and ask for a guideline. if I had proposed them one by one (so what occurs with many other properties in other fields), it should have been fine. I see them all together and I see no difference with many other groups of IDs when you group together, there is no difference in the fact that they come from two people or from random users over two years. That's why I used the verb "calibrate". we really need a zero of the scale here.

Your examples show empty data for very famous authors, we did not request these databases for such profiles but for the less common authors, relevant people whose information are however more scattered on the internet, a case where is very crucial to have clear verified information so we can reference other key metadata of their item in an elegant way. We never even opened these hgher profiles because that's not the goal in ou workflow. You can left those unimported and the database still will fulfill its goal. BTW how much time did it take to you to find those? We can still skip Bompiani.--Alexmar983 (talk) 21:13, 9 May 2019 (UTC)

BTW, IMHO the best is probably the Mondadori ID. It contains more than 2000 summaries, I have not found one yet and with missing text and they contain useful information for referencing.--Alexmar983 (talk) 23:23, 9 May 2019 (UTC)

@Alexmar983: You ask for guidelines but why do we use a process like the property proposal if we have guidelines ? If the creation of a property was just a question of filling some criteria, then no need of the current process of presentation and evaluation.

Your attitude indicates a lack of understanding about what happens here in WD: we can access to hunderds of data sources, but the goal is not to connect everything together but to share knowledge. So for this purpose, it is correct to focus first on the main data sources.

If you want to be successful in your property proposal, you have to have a better selling spirit than what you show until now. I asked about the advantages of your databases, you mentioned data about biography, I just checked 6-8 names in one database and I found 2 completely empty so I assume 20% of useless entries. Ok, please show me I am wrong: execute a kind of query on your databases to show how much entries are empty, or provide other facts like most of the authors in those databases are not present in others databases, give number, type of data, structure, authority of the database,... I am honest: your databases are not sexy for a datascraper like me. If you don't like my opinion, just go away to find another contributor who will buy your poor show. Or directly go tho property proposal and start the process, but at the end, don't come back and complain about poor favorable support: we have other things to do than performing a deep analysis for potential interest on each proposal. You do the proposal, do your job until the end. Snipre (talk) 23:49, 9 May 2019 (UTC)

(confl.)Snipre I understand wikidata they are aspects of wikidata which you are not interested. BTW whatever you think about me, fine. If you would like to change opinion, feel free to contact me. Just one thing: I am not here to sell anything because this is precisely the social dynamics that produce poor databases and poor knowledge of their dynamics, very fragile working environment and disfunctional communities and I want nothing to do with that. I prefer to be disliked or misunderstaood than encouraging them at any level. Wikidata does not stop at getting one thing you want done, that's a very poor vision and I am glad i don't have it, it's a good sign I get this much more than you give me credit. --Alexmar983 (talk) 01:01, 10 May 2019 (UTC)

I had a similar experience as Snipre when I selected several random links at Rizzoli (the only publisher I've heard of). I found two biographies that were empty, only one with dates, and a couple of errors (such as the same book listed under two different authors). The overall impression is that their data is poor. I also question the long-term stability of such data. --EncycloPetey (talk) 00:29, 10 May 2019 (UTC)

if you have time, please cite a "promotional, and very POV" one. that sentence was very strong, it really made me curious. About the rest, as I said myself some are empty, and I pointed out that in such database they are those of very important writers, not the ones of the "less important" writers, the ones we need in the workflow to create more robust items. Some items on wikidata are already linked to databases that contains less information for one entry than another and this was never an issue per se. We can stretch an aspect that we globally ingore in another case, like the superposition cited above, but in the end the question is if they are reliable (they are, they are empty not wrong), their presence increases the quality of the data. You can simply give the precedence to the ones that have no empty descriptions so far 8like Mondadori) and wait for the others more years, and you can still get a good improvement of enough items of writers for the moment. Improvement that we don't get with other databases. VIAF ID are very minimal, they are the only ones we can get, while the BNCF IDs (the National Libraries) are far from complete, without the presence of the main private publishers it's complicated to cover Italian writers.--Alexmar983 (talk) 01:15, 10 May 2019 (UTC)

Marc 21 to Wikidata

Hi everyone. The National Library of Uruguay has around 165.000 entries of edition items that are described on the MARC 21 (Q3919686) format, Before start trying to map that standard to wd properties i wanted to know if anyone recalls if someone did it before of if there is any documentation of that process anywhere. Regards and thanks!--Zeroth (talk) 11:01, 19 January 2019 (UTC)

Notified WikiProject Books participants. --Zeroth (talk) 13:07, 20 January 2019 (UTC)

@Zeroth: There is some mapping from UNIMARC to Wikidata, described here. Kcoyle (talk) 20:20, 18 May 2019 (UTC)

Thank you @Kcoyle:!, i'll check that.--Zeroth (talk) 16:11, 19 May 2019 (UTC)

contemporary constraint

Please comment at Property_talk:P50#contemporary_constraint. This affects many, many editions of books. --EncycloPetey (talk) 03:03, 18 May 2019 (UTC)

Work vs edition

Please help me to understand. If an original work (item A) can have different translations (say, items B/C) then we can consider A as a piece of information (sujet, author, date of creation) independent of language, can't we? Original publication of the work (first, in a language of creation) should have its own item which would contain language of work or name (P407). So I conclude that A should not have language of work or name (P407) and title (P1476). So for example in The Family of the Vourdalak (Q1212326) we should extract all notions about French in a separate item shouldn't we? --Infovarius (talk) 08:08, 3 July 2019 (UTC)

@Infovarius: Your model could work if you are able to track all editions/translations in order to find the first original edition.

To be able to know in which language the first edition (or event the first draft) of a work was written, you need to find the corresponding edition item. How do you want to do that ?

The edition number only allows you to determine the serie of editions by the same publisher. If several publishers in the original language print a work, then you will have several edition number = 1. Same for translation, for each translation and each different publisher publishing it, there is a first edition. So as you have several edition number =1 for the same work you are not able to identify the original edition by that way. The only way is to use the publication date. But then you need to be sure that all editions and especially the original edition has its item. For some reason, a contributor can only create an item about the work and not the corresponding first edition item. In that case how can you apply your model ?

Just an example: for reference reason, I want to add an item about an edition of English work translated in French: I create the item for the translated edition and another one for the work. I don't need the original edition in the original language so I won't spend time to look for the information and for the creation of the corresponding item. By doing that, your model will assume that the language of the mentioned work is French and not English because the only existing edition present in WD is a translated edition. Snipre (talk) 00:14, 4 July 2019 (UTC)

Tool for adding works/versions from OCLC/Google Books

Is there any tool/script to easily create works/versions from an OCLC/Google Books link? This would help a lot with the tedious process of adding items for references. Thanks, Calliopejen1 (talk) 20:02, 17 July 2019 (UTC)

@Calliopejen1: Look at User:MichaelSchoenitzer/CiteTool.js or citoid. Snipre (talk) 05:15, 18 July 2019 (UTC)

@Calliopejen1: CiteTool.js will create a fully-populated reference when you use the OCLC or Google books link as the "reference URL". It does not (as far as I know) actually create items for the work or edition - but I have faith that at some point in the future a bot will be developed that will turn these references into items.

• There is an active project to implement Citoid for Wikidata. If you're interested in following the development effort, here's the Phabricator link. I believe this will handle references/citations; I don't know that it will create items for books/editions.
• If you already use Zotero to keep track of your references, then you might also look at Wikidata:Zotero. It creates QuickStatements input to create an edition-level item. It's frustrating to me in a number of ways. It uses P31=Q571 ('book', the old recommendation, rather than 'version/edition'), and it doesn't look anything up, so you don't get the QIDs for the author (if they exist), publisher, place of publication, distribution format, etc. I seem to recall you have to change "english:" to "en:" as well.
• After experimenting with extremely complex spreadsheets to create work/edition pairs, I have settled on the following process as easier:
  • Create the "work" item using the Cradle book (work) form (yes, you still have to copy/paste a lot of stuff)
  • Duplicate the work item to make the edition, using the "Duplicate this item" tool (script added to your common.js - see User:Magnus_Manske/duplicate_item.js). Add a description, change the P31 to "version/edition", remove the main subject, add publisher, place of publication, etc., and link the work and edition to each other. - PKM (talk) 20:55, 18 July 2019 (UTC)

@PKM, Snipre: Thanks for all this. A bummer that it's such a complicated process! I guess an opportunity for some volunteer coder later on... Calliopejen1 (talk) 21:38, 18 July 2019 (UTC)

How to model manuscripts, books and editions

Hi everyone,

I am working in the item Le Canarien: crónicas francesas de la conquista de Canarias (Q65600504). It is a work (printed book) in which its two authors translated, analyzed and commented the original manuscript, Le Canarien (Q1478385). The first time I thought in how to work with these items and how to model them was: the manuscript, which has a book in which it was translated and this book has an edition, each one with its own Wikidata item. However, I a bit confused. I read about the manuscript properties but I admit you it didn't clarify me a lot.

I still have the next doubt: should I create an item for the book and for the edition? Or, should the book being considered itself as a edition of the manuscript? I am going to leave the items without edit them until I received an answer of you, to not work in vain nor mess up a lot the data.

Thanks in advance for your attention!

Regards, Ivanhercaz (Talk) 00:34, 17 July 2019 (UTC)

@Ivanhercaz: You need an item for the work, one item for the printed edition and one item for the manuscript

Le Canarien: crónicas francesas de la conquista de Canarias (Q65600504) is the printed edition of the work defined by Le Canarien (Q1478385)

An additional item is necessary to described one of the two manuscripts. If you create that new item, you have to add the following properties : instance of manuscript and exemplar of Le Canarien (Q1478385).

Simple advice: never use the term book when describing your problem in WD. Book can be the work, the edition of the exemplar depending on the point of view. Snipre (talk) 12:41, 17 July 2019 (UTC)

To avoid mistake I created two items, one for each existing manuscript: codex Montruffet (Q65617317) and codex Egerton 2709 (Q65617336). Snipre (talk) 13:01, 17 July 2019 (UTC)

@Snipre: Thank you very much for the explanation! And for the creation of the items too. I am going to make my own scheme to keep this in mind for the next time in which I need to address a similar situation. Regards, Ivanhercaz (Talk) 16:47, 19 July 2019 (UTC)

Idea of cleanup

Hi,

We decided not to use book (Q571) a long time ago but there still is a lot of item using it (68238 right now according to this query https://w.wiki/6hs). Shouldn't we replace this value by written work (Q47461344)? Maybe not in all items (we may need to do some double-checking before) but at least most of them I guess.

WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.

What do you think? Is it a good idea?

Cheers, VIGNERON (talk) 09:58, 1 August 2019 (UTC)

I would support that (with some temporary qualifier for future checking). Book is "false" in any case and written work would be mostly correct.

Perhaps we could also check manually in the first place the book published after the author's death (which are more likely to be about a specific edition).

Alexander Doria (talk) 10:35, 1 August 2019 (UTC)

I would support a cleanup to remove book (Q571), but making them all written work (Q47461344) provides no useful information, and could easily be wrong. We need to know first of all whether it is a data item for a work or an edition/translation. Second, we need to know what sort of "book" it is: if it is a novel (Q8261) or a scholarly work (Q55915575) or something else. There are already many editors here who think that written work (Q47461344) is the correct value over all others, even at the expense of clarity or specificity. We should not promote that idea by a blanket change. Surely our tools are better than that by now. --EncycloPetey (talk) 15:12, 1 August 2019 (UTC)

@EncycloPetey: true, how do you suggest to proceed? (I have several idea but I'd love to hear other point of view). Cheers, VIGNERON (talk) 16:06, 2 August 2019 (UTC)

Any of the items tagged with a value for translator (P655) or editor (P98) or publisher (P123) is a candidate for version, edition or translation (Q3331189). I would start with those, and then recount the remaining items to see what progress was made. --EncycloPetey (talk) 16:19, 2 August 2019 (UTC)

A combinaison of instance of book (Q571) + one of the dedicated properties of edition has to lead to the instance of written work (Q47461344) definition. Not sure that property translator (P655) is the best one to identify version, edition or translation (Q3331189): some works are mixing different languages. The most important properties to identify recent editions are edition number (P393), place of publication (P291), publisher (P123) or publication date (P577) in my opinion. Snipre (talk) 22:39, 5 August 2019 (UTC)

publication date (P577) isn't a dedicated edition property; it is used on works as well as editions. It is used to indicate the date of first publication of a work. I'd also point out that I've come across more than a few works which had edition data included on them (e.g. ISBN) when that data was imported from a Wikipedia page about the work. --EncycloPetey (talk) 23:34, 5 August 2019 (UTC)

@EncycloPetey: According to data model defined on Wikidata:WikiProject_Books, publication date (P577) is an edition property. For me this is the only truth: the data model has to be respected unless it explicitely mention some exception. So the question is not to know how people are using the propeties in a wrong way, the question is to define if publication date (P577) should be an edition property only or should be extended to work. Once this done then we will have to clean the items according to data model. Snipre (talk) 05:55, 6 August 2019 (UTC)

And which should be a date for a work? inception (P571)? --Infovarius (talk) 10:55, 7 August 2019 (UTC)

@Infovarius: Can a date represent the writing of a work ? But yes, inception (P571) is a better property for work (this is the way mentioned in the data model) but I would be really happy if someone can explain the use of that property in this scenario: is the date the date of the beginning of the writing or the end of it ? An author can start to write in a period of his life and then stop a moment to finish later, how can we model that ? Again trying to represent the writing process with one date is a huge simplification. Snipre (talk) 13:21, 7 August 2019 (UTC)

The problem with inception (P571) is that it might refer to the date an author began writing a work, rather than the date of first publication. As Snipre has pointed out, inception (P571) is not very clear what it refers to. It is really intended to items that have a clear beginning (and ending) date, such as the existence of a company or nation. --EncycloPetey (talk) 16:27, 7 August 2019 (UTC)

@EncycloPetey: For me work item should not have any date unless a clear source mentions the writing period of a work. Instead of starting to put edition data in work item, the correct way is to create the related edition item and to add the information at the correct place. I don't understand why the publication date has to be mentioned in the work item. If you want to have that information just do the job correctly: create the corresponding edition item. Snipre (talk) 20:21, 7 August 2019 (UTC)

@Alexander Doria, EncycloPetey, Snipre, Infovarius: this discussion is very interresting but it's going sideways. So if I understand correctly the collective point of view, for items that might be about editions, you prefer to keep book (Q571), ok, no problem. Can I at least do the replacement book (Q571) -> written work (Q47461344) on the 22167 items that are most likely not editions https://w.wiki/6ry (by removing all the items with any of these properties: translator (P655), editor (P98), publisher (P123), illustrator (P110), author of foreword (P2679), author of afterword (P2680), publication date (P577), place of publication (P291), ISBN-10 (P957), ISBN-13 (P212)), and I probably won't touch the others since it's too much of a mess (a lot of them seems to be merge of several editions/works/items/whatever :/ ). Cdlt, VIGNERON (talk) 14:53, 7 August 2019 (UTC)

However if the work is a novel (Q8261), then that value should be used instead of written work (Q47461344). We've had this discussion several times now, and have repeatedly come to that conclusion. written work (Q47461344) is only the most general possible value, and if a more precise value exists, we should use it. --EncycloPetey (talk) 15:43, 7 August 2019 (UTC)

Another suggestion: If book (Q571) is not the only value for instance of (P31) on an item, then book (Q571) can simply be removed, rather than replaced. I've poked around and found a few cases where there were multiple values present. --EncycloPetey (talk) 19:48, 7 August 2019 (UTC)

@EncycloPetey: No, we never agree that a work can be a instance of everything: a work can be an instance of a subclass of written work. Novel is not a subclass of written work but an instance of literary form. So instances of novel can't be a work. Snipre (talk) 20:14, 7 August 2019 (UTC)

@Snipre:: novel (Q8261) used to be a subclass of literary work (Q7725634). There is no separate property for literary form approved for use by this community. The form is the value for "instance of". And we have agreed as a group (several times, including one instance still visible above) that instance of (P31) should be novel (Q8261) for novels, because this aligns with the FaBio tree at Wikidata:WikiProject Books/Works. --EncycloPetey (talk) 20:55, 7 August 2019 (UTC)

@VIGNERON: You can do the inverse: if an item with instance of book contains 3 or more properties related to edition class, then we can define it as instance of edition. Snipre (talk) 20:17, 7 August 2019 (UTC)

@EncycloPetey: isn't written work (Q47461344)novel (Q8261) a value for genre (P136) and not instance of (P31)? (it's how it's used 90 % of the time). That said, the remark is interresting and I checked: there is 122 items with both book (Q571) and novel (Q8261) as value for instance of (P31), then I guess we could remove the book (Q571), replace it by written work (Q47461344) and move written work (Q47461344)novel (Q8261) in genre (P136)). For the other cases of instance of (P31) with book (Q571) and something else, there is 3208 items (again quite low) and most of the time, the second value is wrong: https://w.wiki/6sm

@Snipre: I thought about it but I guess I can't because the inverse of "not once this property" is "any number of value for this property" so if there is multiple values, then it's not *one* edition, it's multiple editions (which is a whole different problem that can't be treated automatically - because sometimes one edition has actually several values -, or maybe it's possible but I don't see how, I looked at the data but it seems to be quite a mess). Bottomline: I'm just aiming at the easy low hanging fruit there. Cheers, VIGNERON (talk) 21:02, 7 August 2019 (UTC)

No, "written work" is not a genre. If it is being used that way, then I suspect someone ran a bot replacement incorrectly. --EncycloPetey (talk) 21:05, 7 August 2019 (UTC)

Obviously bad copy-paste, corrected, VIGNERON (talk) 06:31, 8 August 2019 (UTC)

No, novel is not a genre. People have been putting that in because it was in our data model, but we have repeatedly agreed that novel is not a genre. There is a discussion further up this page where we agreed so. We keep having this discussion, but each time we have it, some people seem to have forgotten that we decided novel was a literary form (like poem, play, or scholarly article) rather than a genre (like Gothic literature, Western literature, historical fiction). Novel is a property of the length of a work of fiction, and not a property of its style or content. Genre describes style and content, not length. It is ridiculous to say that War and Peace (Q161531), Gulliver's Travels (Q181488), and The Shining (Q470937) are all in the same genre. They are all novels (long prose works), but they are in very different genres. --14:10, 8 August 2019 (UTC)

@VIGNERON: I agree with EncycloPetey: if novel is not a genre, how do you define classification based on content (SF, erotic, historical,...) ? We need 2 different properties to describe a text according to its content or its form. Snipre (talk) 10:04, 14 August 2019 (UTC)

short story collection (Q1279564) is also often used with genre, even though it violates a constraint. Likewise, short story (Q49084), which has been declared to be a genre so doesn't violate the constraint. Ghouston (talk) 01:45, 15 August 2019 (UTC)

RfC about model for document description

Dear all,

I started a RfC to improve our model for document description. As I am not all-knowing, I did a proposition of model and I expect some modifications before we can launch a decision phase. The principle of that 2 phases RfC is to avoid to start the decision before having input from a large number of contributors mentioning particular cases which can change the original proposition. This is always frustrating to give its opinion about a proposition which changes several times between the start and the end of the decision phase.

Please have a look at Wikidata:Requests for comment/Improved model for document description. Snipre (talk) 13:16, 14 August 2019 (UTC)

@VIGNERON, EncycloPetey, Alexander Doria, Andreasmperu, Pasleim, TomT0m: @Infovarius, Freddy eduardo, Aubrey, Micru, Tpt, Kolja21: Snipre (talk) 13:19, 14 August 2019 (UTC) @Hsarrazin, PKM, Giftzwerg 88, Jura1, Beat Estermann, Jc3s5h: Snipre (talk) 13:22, 14 August 2019 (UTC)

Novel used with instance of (P31) instead of genre (P136)

@EncycloPetey: believes that there’s a consensus to use genres as a statement for instance of (P31) instead of genre (P136). Given that I failed to find that specific discussion, and that I haven’t bumped into anybody else doing this type of edits, I think a clarification is needed. --Andreasm ^{háblame / just talk to me} 23:32, 3 March 2019 (UTC)

Incorrect. I have never advocated the use of genres as a statement for instance of (P31). What I have advocated is the use of novel (Q8261), which is a subclass of fiction literature (Q38072107), which in turn is a subclass of literary work (Q7725634). This is also laid out clearly in the FABiO chart at Wikidata:WikiProject Books/Works, as we discussed at length, where again "novel" is listed as a subclass of "literary work". --EncycloPetey (talk) 23:53, 3 March 2019 (UTC)

I assume the confusion arises because novel (Q8261) is both a subclass of literary work (Q7725634) and instance of genre (Q483394). Although both statements are correct, the data structure would become clearer if we restrict ourselves to one of the two variants. --Pasleim (talk) 10:09, 4 March 2019 (UTC)

From WP-EN: "A novel is a relatively long work of narrative fiction, normally written in prose form, and which is typically published as a book". This definition shows the problem with the use of novel as a literary genre:

• "a relatively long work": "long" here is a format (physical) attribute, not a conceptual one. novella (Q149537), novelette (Q472808) and short story (Q49084) differ form novel (Q8261) only by the size (measured in number of words) of the literary work, not "thematically". They all are "narrative fiction, normally written in prose".
• "typically published as a book": not really, there are a lot of novels (for example many science fiction novels) first published as serializations in magazines, and not every serialized novel has been later published as a book. The confusion between "book" and "novel" is what lies at the heart of this discussion, and the point is that books and novels are completely separate things: not every novel is published as a book, and therefore novels are not a subset of books.

Take for example a well-known novel such as Fahrenheit 451 (Q202009). It has been classified as philosophical novel (Q2016518) and dystopian novel (Q26928598). But what about The Ones Who Walk Away from Omelas (Q7755107)? That's also a philosophical fiction, but it's not a novel (Q8261) but a short story (Q49084). So we have to create a "philosophical short story" item, ok? And then we also have to create "cyberpunk novelette", "mystery novella", "science fiction short story", "distopian novelette", etc, etc. We have now at least M x N "genres" from the combination of (fantasy, mistery, dystopia, ...) with (novel, novella, novelette, short story). It's possible to do this? Yes. Is it appropriate? It doesn't sound as a good ontology (to me at least), it does sound like we are trying to blend two classification criteria into one. --JavierCantero (talk) 10:53, 4 March 2019 (UTC)

novel (Q8261) can't be a genre: genre is about the content and not about the format. In reality, books can be described by several parameters:

• genre: classification of the content of a book (science-fiction, fantasy, biography, ...)
• form: classification of the text format (novel, short text, roman,...)
• support: classification according to support structure of the text (scroll, codex, tablet, hardcover, papercover...)
• material: classification according to the material (papyrus, parchment, paper, vellum...)
• writing type: manuscrit, printed,

So we can have a science-fiction novel printed on a papyrus scroll, an erotic roman hand written on a parchment codex. we definitively need 5 properties to describe correctly a book. Snipre (talk) 20:11, 4 March 2019 (UTC)

In practice, only the first two are used to describe the work; the latter three describe an edition or manifestation only. A science-fiction novel will be a science-fiction novel regardless of what it is written upon or the writing type used to encode the text. Only the form and genre will be invariant among all editions, and therefore edition independent. --EncycloPetey (talk) 21:39, 4 March 2019 (UTC)

novel (Q8261) shouldn't be used as a genre, but the example value for genre (P136) in the project page says otherwise. And it seems that at least the OP also thinks that way. --JavierCantero (talk) 23:07, 4 March 2019 (UTC)

The "OP"? --EncycloPetey (talk) 23:42, 4 March 2019 (UTC)

The example mentioned is The Hours (Q2610868), which is currently identified as an instance of written work (Q47461344) change made Nov 2018, and has four different values for genre (P136): parallel novel (Q6619804), feminist novel (Q26987767), LGBT (Q17884), novel (Q8261). The last (novel) was added when this merge was done in Dec 2017. --EncycloPetey (talk) 23:49, 4 March 2019 (UTC)

(OP = "Original Post", commonly used to refer to the comment that starts a discussion thread) --JavierCantero (talk) 11:41, 5 March 2019 (UTC)

I’d take an opposite direction because forms and genre, in my humble opinion, can be totally mixed. For example « comic strip » is something caracterized as a specific form and usually a humor genre, but it’s definitely something on its own. I’d definitely use instance of (P31) to characterize xkcd as a comic strip for example, this emcompass both some kind of genre and a kind of form. This can coexist with other properties but in the end the generic classification properties are really useful to model these « blurry » situation which are, in my opinion, more like the norm than the exception. And everybody knows what a comic-strip is, so it’s useful. I’d use metaclassification to classify items such as « comic-strip », for example

⟨ comic-strip ⟩ instance of (P31) ⟨ form ⟩

and

⟨ comic-strip ⟩ instance of (P31) ⟨ genre ⟩

, and benefit the flexibility of the generic classification properties while fully maintaining precision. Which is not easy to do as these objectives may seem opposite.

It would also be possible to use constructions such as

⟨ comic-strip ⟩ has quality Search ⟨ Q47524457 ⟩
form Search ⟨ comic ⟩
genre (P136) ⟨ humor ⟩

or stuffs like that. author  TomT0m / talk page 11:30, 14 August 2019 (UTC)

@TomT0m: Your example is a particular one and I would prefer to start to build a model based on the general case. Do we agree that a text can be classified by several characteristics, at least by genre, but other exists ? If yes, then instead of focusing on particular cases, can we define the different characteristics of a text and do we agree to create a property for each of them ? That's the basics. Snipre (talk) 11:40, 14 August 2019 (UTC)

I don’t think that’s really a fruitful approach, as this problematics is transversal to artworks in general, (forms and genre at least), so before focusing on what is specific to text it’s useful to ask ourselves « how do we classify artworks ? » (this is where the flexibility of the generic properties allow to easily capture common type of art defined by a variety of criteria that nevertheless have a name). When these broad classes are defined we can focus on what’s missing and which properties are used to define the classes (the length of a text …) . But takes items like Burlesque poetry (Q3401086). Clearly, an artwork of that type is a poem of a certain genre, we can superclass it with both. If we say that it’s an instance of Burlesque poetry (Q3401086) we already said a lot at a cheap cost. author  TomT0m / talk page 11:56, 14 August 2019 (UTC)

@TomT0m: The problem of that kind of classification is the lack of systematic: do we have an item for all types of poetry ? Do we want to create similar items for book, movie, theatre, TV serie,... ? Again, if we want to be efficient and thinking to data extraction by machine, we should be more programming oriented. Snipre (talk) 13:05, 14 August 2019 (UTC)

@Snipre: We don’t have to create all those items, that’s the beauty of the stuff. It’s enough to create two « instance of » statements to obtain exactly the same result than two statement « form + genre ». Search for all burlesque poetry ? query ?poem wdt:instance of/wdt:subclass of* <burlesque> ; ?poem wdt:instance of/wdt:subclass of* <poem>. You’ll also find the instances of « burlesque poetry ». author  TomT0m / talk page 13:23, 14 August 2019 (UTC)

@TomT0m:, of course we can put all properties into some P31 or P279, but should we?.. --Infovarius (talk) 10:14, 15 August 2019 (UTC)

@infovarius: We can't, for a start :) Datatype properties can't naturally, for example, date of birth and so on. Properties like parenthood and so on will never fit neatly either. But here, I have the feeling, backed up by a really quick google search https://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780199279456.001.0001/oxfordhb-9780199279456-e-8 (a quote of the page, There are a number of traditional classifications of the arts, for instance in terms of their media (stone, words, sounds, paint, etc.), their species (sculpture, literature, music, drama, ballet, etc.), or their styles or contents (tragedy, comedy, surrealism, impressionism, etc.). The ontology of works of art does not map neatly on to these classifications, however." I did not read the book) that the artwork classification is something that actually needs flexibility and is transversal to a lot of criteria. It's really a classification matter. A genre like romantic comedy (Q860626) clearly defines a subtype of film.

The genre (P136) property should nevertheless be useful, even if we chose to not use it on artwork themselves, for example to link a music band to the kind of music it creates, for example. This avoids a total parallel explicit classification of artists and artworks. author  TomT0m / talk page 13:55, 15 August 2019 (UTC)

EDIT : the english article also says romantic comedy can also be a play. author  TomT0m / talk page 13:59, 15 August 2019 (UTC)

linking book review item to book author item and literary work item

Here's:

• an author: Bonnie Effros (Q55620197)
• her book (P31=literary work) Merovingian Mortuary Archaeology and the Making of the Early Middle Ages (Q56284851)
• a review of her book Bonnie Effros, Merovingian Mortuary Archaeology and the Making of the Early Middle Ages: the Transformation of the Classical Heritage. (Berkeley: University of California Press, 2003, xviii + 272 pp., 22 ills, ISBN 978 0 520 23244 0) (Q57888271)

How should the review of her book item be linked to her item and to the book item? (Not sure which properties to use). thx --Tagishsimon (talk) 21:45, 19 August 2019 (UTC)

@Tagishsimon: Have a look at review of (P6977). No need to linked the review to the author: this done by the links between the review/work and work/author. Why to add a third link review/author ? Snipre (talk) 08:22, 20 August 2019 (UTC)

Excellent, Snipre; thank you. --Tagishsimon (talk) 15:29, 20 August 2019 (UTC)

Many bad Cewbot editds

It looks as though @Cewbot: is adding lots and lots of bad data (e.g. [1]) to numerous data items.

There are two problems: (1) no book has 0±0 pages, as that is nonsense. (2) all the data about page numbers, illustrators, etc. are being added to data items for works rather than specific editions. --EncycloPetey (talk) 02:07, 8 September 2019 (UTC)

Work with one edition

Perhaps a very stupid question..... What do I do with a written work, that has only one edition? Do I have to create two wikidata-items? And what are the naming conventions for those two items? (see e.g. Q21962241) --Dick Bos (talk) 17:50, 2 October 2019 (UTC)

@Dick Bos: Can you ensure that no new edition will be edited in the future ? I think no, so two items is recommended. For the name, both items can have the same label. For edition there is a long unresolved discussion if it is a good practice to add the edition number in the label of edition item. Ex.: blablabla (1st ed.). And in the description, use work or edition and not book. You are welcome. Snipre (talk) 11:47, 3 October 2019 (UTC)

Publication date of work

There is a concept to describe this section : an evergreen (Q55643019). So I am aware of the comments above describing a data model agreement on Wikidata:WikiProject_Books, which state that a publication date (P577) is not a work property but only an edition one. Yet no model shall indefinetly be accepted, and evolution can, at least, be discussed. So I continue on the sideways of the clean up section above. I would humbly claim that a publication date (P577) is adding information to a work, even though it might be duplicated information with an edition. More precisely, it seems relevant to me to know when the first edition of a work has been published (avoiding metaphysics of when a work is created/written/incepted). So we actually could reduce it to the publication date (P577) of a work. If critics say this property would then be too broadly used, at least a property "first edition date" shall exist (naming is propably not the best, but you got the idea : having any kind of date information on a work). Jumtist (talk) 22:35, 22 October 2019 (UTC)

I think i debate with my self here, since publication date (P577) now states "date or point in time when a work was first published or released" which is exactly what I proposed... Jumtist (talk) 17:38, 24 October 2019 (UTC)

Which would imply that it shouldn't be used on editions? Perhaps the description could be tweaked a bit. Ghouston (talk) 22:18, 24 October 2019 (UTC)

In my opinion, a "first publication date" property for written work (Q47461344) items could help to assess its copyright status (whether it is in the certain country's public domain or not) so it could be useful to have it. It's also common to user the notation title (year of publication) to refer to creative works (including written works), so to be able to retrieve this info from Wikidata through a standard template could be a possible use case to consider. --JavierCantero (talk) 09:33, 3 November 2019 (UTC)

Interesting question: for manuscripts, should publication date (P577) give the date of first creation, or the date of the editio princeps (Q1249682), the first printed edition? As Javier notes, printed publication can be an important issue for determining copyright status, eg in the UK where copyright in unpublished manuscripts is going to persist until 2039, regardless of how long ago the author died.

Presumably this should be the case for how we use publication date (P577), even if copies of a manuscript were widely made and circulated in medieval times. But, to take an extreme case, does this mean that for Bible (Q1845), or for Vulgate (Q131175), that publication date (P577) should equal the date of Gutenberg Bible (Q158075) ? Jheald (talk) 10:21, 3 November 2019 (UTC)

@Jheald: The model indicates inception (P571). A manuscript is not published (no printing process). If a work was copied several times by hand then we have an edition but with a publication date lasting several months or years, and each individual copy has an inception date. Bible (Q1845) is not a work as the content is changing depending on the publisher/editor/religious group. There is no unique version of the Bible. Snipre (talk) 08:00, 5 November 2019 (UTC)

External identifiers

Hi All,

I wrote a proposal about expanding the statements regarding external identifiers, please have a look and add your opinions: https://www.wikidata.org/wiki/Wikidata:Project_chat#External_Identifiers_-_expanding_statements,_best_practice --Adam Harangozó (talk) 13:27, 4 November 2019 (UTC)

Blog post about a small Wikisource pilot with Structured Data on Commons

Hi all! To explore the potential of Structured Data on Commons and Wikidata for Wikisource, Satdeep and a few members of the Punjabi Wikisource community did a small pilot project with a set of Punjabi books. You can read more about the pilot in this blog post: https://space.wmflabs.org/2019/12/07/how-can-structured-data-on-commons-wikidata-and-wikisource-walk-hand-in-hand-a-pilot-project-with-punjabi-qisse/ and its project page. We would very much welcome your input in the data modeling aspects of this pilot, and hear your thoughts in general, also if you have ideas on better Wikisource / Wikidata / Commons integration. Many greetings, SandraF (WMF) (talk) 14:16, 9 December 2019 (UTC)

Is this supposed to happen

When making items about the different editions of Islam and Reconciliation (Q77826862) i could not help but notice that both OCLC 770816192 and 144591218 share the same ISBN-10 and ISBN-13. Is this normal? And should both OCLC entries get their own items? --Trade (talk) 00:44, 10 December 2019 (UTC)

It's not supposed to happen, no, but WorldCat does contain a lot of errors. I would not base the creation of data items upon OCLC listing. Rather, if two OCLC items appear to be the same thing, list them both on a single data item. Personally, I omit OCLC from items for specific editions. There are just too many errors. --EncycloPetey (talk) 01:32, 10 December 2019 (UTC)

@Trade: When there is an error in an external database, it is a good practice to contact the database administration and to mention the error: sometimes there is a positive feedback and a correction of the data. Snipre (talk) 07:58, 10 December 2019 (UTC)

@Snipre:, any idea what the email of OCLC is? @EncycloPetey:, what do you use when creating items for specific book editions? --Trade (talk) 23:41, 11 December 2019 (UTC)

Most often I am working from copies hosted somewhere, especially Wikisource. This means I can pull information from a scan of the work directly, and can verify data against a scan of a volume. To find the scans, I use Internet Archive, Hathi Trust, the Biodiversity Heritage Library, or another well-curated collection of quality scans. --EncycloPetey (talk) 00:00, 12 December 2019 (UTC)