Wikidata talk:WikiProject Books/2022

Should the title of a book be translated for its label in other languages?

Is there a convention for translating the title of a book to other languages for using it in those languages' labels? Here's an example so that I get my point clear: Suppose there's a book that is published with its title in Spanish. I know the Spanish label of this book should be the title, but my question is: Should the label in other languages be (1) the translation of the title in those languages or (2) the original title?

Rdrg109 (talk) 04:16, 5 January 2022 (UTC) (please ping on reply)

I would say only of if it has been translated in reliable sources (official translation published - sometimes there are several of them btw), or title translated in some publications whereas the book is not translated (for example, Q7732034 has (unfortunately) not been translated to English, but there are some articles which list the English name). Ymblanter (talk) 19:48, 6 January 2022 (UTC)

Are you asking about the work item for the book, the editions of that book, the instances of the book, or something else? I would not give the same answer to each question. For major works, most national library databases will have a translated title for the work database record, but they usually will not translate the title of a specific edition. For a work data item here, I would give the local language title (if one exists), but include the original title (if known) provided that it's written in the same alphabet or a transcription system suitable for that language as used by the national database from the country that uses that language. --EncycloPetey (talk) 22:32, 6 January 2022 (UTC)

Example of an article in German: The opus fantasy. Fantasied work as a metafantasy in the creative process (Q33473649). The English translation is added in square brackets. Imported from PubMed (Q180686). --Kolja21 (talk) 00:34, 8 January 2022 (UTC)

+1 with both Ymblanter and EncycloPetey. If by book you mean work, then it *can* be translated (providing sources), if you mean edition it probably shouldn't be translated. Plus, label and title shouldn't be confused (it's two different things, and we have title (P1476) - and others - for the "true" title so we can allow some latitude and flexibility in the labels). Of course, as always with bibliographic data, there can be exceptions and complications (indeed transcription and transliteration are an other can of worms, but that's beside the original question here).

@Kolja21: interresting case, this is maybe and kind of acceptable as a degraded solution but ideally it should be replaced with the original title in German ("Die Opus-Phantasie: Das phantasierte Werk als Metaphantasie im kreativen Prozeß" apparently according to https://elibrary.klett-cotta.de/article/99.120105/ps-33-3-193 ).

Cheers, VIGNERON (talk) 16:27, 2 February 2022 (UTC)

How to deal with multiples ISBN on items?

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

Hi,

Tough question (that we talked about already but never answered), not sure we can have an answer but I hope some cataloguing aficionados and wikidatian modeler could help.

First, on Wikidata we have a lot of items with multiples ISBN. This simple query https://w.wiki/4nJW gives 4872 results right now (and only for ISBN-13 (P212)).

Most are errors that need to be corrected (including but not milited to: splitting the item in two items for each editions, removing the ISBNs if they don't belong on the item and/or put the rank to deprecated if one the ISBN was printed wrong) but then, in rare case, some specific editions indeed have multiples ISBN (like Q40220865, Les manuscrits à peintures en France (Q109671685) or Dix mille saints (Q108369181)).

This could be very confusing (and external catalog are sometimes confused). Could we imagine a way to indicated that these cases are not errors? (for instance with a qualifier?)

Cheers, VIGNERON (talk) 16:55, 2 February 2022 (UTC)

Imho adding a qualifier like publisher (P123) should be enough. It's the same problem with authors and authority control. We use qualifiers like subject named as (P1810), indicating that the second ID is not a duplicate, but lists like Wikidata:Database reports/Constraint violations/P227#"Single value" violations are unable to differentiate. --Kolja21 (talk) 19:12, 2 February 2022 (UTC)

@Kolja21: adding publisher (P123) in qualifier is enough to remove the constraint but not enough to differentiate multiple values by error and multiple values on purpose. My iea is to have something to make the difference so we can built system (queries or otherwise) to have a list of "true" errors without false-positive. More importantly it would let the user know this is on purpose. But following your idea, I refined the query to only list items with ISBN with publisher in qualifier: https://w.wiki/4nMF Now we have less results (I'll clean them soon) but there is still both error and correct cases mixed in these results. Cheers, VIGNERON (talk) 20:30, 2 February 2022 (UTC)

cover image

Is there a way to indicate cover image of a book? For example https://www.wikidata.org/wiki/Q110934139 has the cover image https://www.wikidata.org/wiki/Q19883758, but I can't see a way to connect them. P736 cover art by doesn't seem suitable as the cover was designed by a person who used and referenced that particular image. Thanks HelsKRW (talk) 10:18, 16 February 2022 (UTC)

A cover image ought to be the cover of that edition, not a work of art that is included in the cover image. If the cover of the book is not copyrighted, and if you can extract the cover image for the book, that can be used as an image (P18) for the edition with object has role (P3831) as a qualifier. However, in this case, it seems the original is copyrighted, so we wouldn't be able to use it or the derivative cover. The author/publisher may be required to change the cover. --EncycloPetey (talk) 17:21, 16 February 2022 (UTC)

Property proposals

• Wikidata:Property_proposal/Databazeknih.cz_Book_ID
• Wikidata:Property_proposal/Databazeknih.cz_Author_ID JAn Dudík (talk) 20:22, 16 February 2022 (UTC)
• Wikidata:Property proposal/CBDB.cz Author ID
• Wikidata:Property proposal/CBDB.cz Book ID --Lukša (talk) 09:30, 17 February 2022 (UTC)

i'd also add Wikidata:Property proposal/літературний редактор to a list of related proposals.Vojtěch Dostál (talk) 08:33, 23 February 2022 (UTC)

What about book (Q571)?

I miss some guidance on how book (Q571) should or shouldn't be used. Property_talk:P31#None-of_constraint:_book? mentions that it's problematic or at least disputed. --Jonas kork (talk) 08:29, 26 February 2022 (UTC)

Hi @Jonas kork: don't use it, this item is ill-defined and problematic. You can use literary work (Q7725634), version, edition or translation (Q3331189) or individual copy of a book (Q53731850) instead. CaLéValab (talk) 18:39, 4 March 2022 (UTC)

Alright, no problem. I suggest someone (who is minimally more familiar with this than me) add a note to the project page. --Jonas kork (talk) 15:07, 6 March 2022 (UTC)

There is already a note in the discussion page for book (Q571). Where on this project's page would you place it? The project page is about the project, not a collection point for discussions (which occur here). --EncycloPetey (talk) 17:53, 6 March 2022 (UTC)

The project has "Books" in the title. The item carrying that label might be seen as an obvious object for this WikiProject. Since Q571 is contentious, I think it's helpful to give visitors and newcomers a heads-up about the problem. I only stumbled across it by accident and assumed this would be good place to find more guidance. To be honest I don't check all discussion pages for items (or properties) that I put into statements. --Jonas kork (talk) 14:15, 11 March 2022 (UTC)

Should every book be split into two items (work and edition)?

I'm familiar with a similar two-fold structure from a different subject/ database, but I'm not sure what the general stance on Wikidata is. If a book was published only once, in a single language (at least so far/ that we know of), would we just need a single item = instance of version, edition or translation (Q3331189)?

Side note: It's unfortunate how the project page talks about the 4 levels of FRBR, and how Wikidata combines two of them, and then goes to sum up: "Thus a two-layer framework (...) has been used." It would be better to mention that the FRBR "item" level has been dropped, or is of no concern in the following description. --Jonas kork (talk) 08:37, 26 February 2022 (UTC)

Hi @Jonas kork: as of today, lots of "work" items don't have associated "edition" items, and lots of "edition" items don't have an associated "work" item. From a pure database point of view, the best would be to have both a "work" and an "edition" item for each published book, but that would be a lot of work for maybe no tangible upside (to be debated). At least, each item should be correctly linked to the wikisource and wikipedia articles, either directly, or indirectly through the associated "work" item or "edition" items, thanks to the edition or translation of (P629) or has edition or translation (P747) properties. CaLéValab (talk) 18:49, 4 March 2022 (UTC)

To elaborate, if there is a Wikisource copy, but nothing on Wikipedia, there may only be an "edition" data item, and if there is a Wikipedia article, but nothing on Wikisource, there is likely a "work" data item. If it exists in both locations, or if there is reason to be coordinating multiple editions, then there should definitely be a work data item and a data item for each edition. As CaLéValab says, sometimes the practicality is that no one has had the time to ensure both things exist and are properly separate with only data apropriate to the type of data item. --EncycloPetey (talk) 05:35, 5 March 2022 (UTC)

Thanks, makes sense. I think it would be helpful if this clarification found its way onto to project page. --Jonas kork (talk) 15:08, 6 March 2022 (UTC)

Requesting your thoughts

Hello all and thank you for the work on this project!

I'm helping to prepare some blog content which will include comments about this project. I'm curious to hear how project contributors think on a couple of points (below). If you have thoughts on these please do share here or leave a comment on my talk page.

• Is providing access to online representations an important motivation for you in participating in this work?
  • Noting the inclusion of properties like Google Books ID (P675), Internet Archive ID (P724), HathiTrust ID (P1844), ...
• Have the Project's suggestions for modeling books been reflected in any EntitySchemas, or is this work planned (or desired)?

Riesengrey (talk) 18:48, 15 March 2022 (UTC)

CC Tructho

"as told to"

Is there an existing qualifier, property, or other method to indicate an author/contributor's role is "as told to"? E.g. Tibet is My Country, by Tubten Jigme Norbu (Q1385565), as told to Heinrich Harrer (Q84211). Or The Big Love (Q7717703), by Florence Aadland (Q111175143), as told to Tedd Thomey (Q16028255). Or The Trials of Jimmy Hoffa: An Autobiography, as told to Donald I. Rogers (Q111175352). Two names generally credited on the book cover, one the originator, the other the recorder/interviewer/compiler/editor. -Animalparty (talk) 05:50, 10 March 2022 (UTC)

The description would be scribe (Q916292) or amanuensis (Q499134) but I am not sure how you would record that a person held that role in relation to a work on that work's data item. editor (P98) is the closest property besides author. --EncycloPetey (talk) 06:42, 10 March 2022 (UTC)

@Animalparty, EncycloPetey: I would have thought that author (P50) will often be appropriate, as they were the person who actually wrote the book. But the generic contributor to the creative work or subject (P767) is also available, if you believe they only had secondary responsibility for the text. As for explanatory qualifiers, object has role (P3831) = ghostwriter (Q623386) or alternatively interviewer (Q46034607) (or both) might be appropriate; perhaps together with object has role (P3831) = interviewee (Q55534929) for the nominal author. Jheald (talk) 13:17, 16 March 2022 (UTC)

form of creative work (P7937)

What did Wikiproject Books do before the existence of this property? Lectrician1 (talk) 16:24, 5 March 2022 (UTC)

Argued, mostly. --EncycloPetey (talk) 04:30, 2 April 2022 (UTC)

What value should I use for edition or translation of (P629) in editions?

Hello everyone! I'm new in the domain of books in Wikidata. I have a question.

Suppose a book has 5 editions. Each of the editions have unique values for ISBN-10 and ISBN-13. This is how I would handle this scenario: I would create an item for each of the editions and I would add instance of (P31)version, edition or translation (Q3331189) and use the properties ISBN-10 (P957) and ISBN-13 (P212) in each of those 5 items. Since they are editions, I would also add edition or translation of (P629), but I'm wondering:

• What value should I use for the property edition or translation of (P629) in those 5 editions?
  • Should I create a Wikidata item for the concept of the book itself that has been improved over the course of the editions?
    • If so, what value should I use for instance of (P31) for that newly created item?

Rdrg109 (talk) 01:18, 2 April 2022 (UTC)

We call the main item for the concept a "work", and yes there should be a data item for the work. For "books" the value of instance of (P31) will usually be literary work (Q7725634) if it is fiction, but it could be a scholarly work (Q55915575) if it is an expert nonfiction "book", or musical work/composition (Q105543609) if the "book" is a musical composition. There are some other values, and it really depends on the kind of "book" that is meant. The term "book" is very vague. --EncycloPetey (talk) 04:29, 2 April 2022 (UTC)

Should we add multiple title (P1476) statements for a work that has multiple translations?

Suppose a written work (Q47461344) is published as the 1st edition by the author in English (Q1860), that 1st edition is translated into Portuguese (Q5146), French (Q150) and Russian (Q7737) and each of the translations is assigned a different title in the target language. Under the guidelines of WikiProject Books, we would now have 5 Wikidata items

1. The work which has the statement instance of (P31)written work (Q47461344)
2. The 1st edition in English (Q1860)
3. The translation of the 1st edition in Portuguese (Q5146)
4. The translation of the 1st edition in French (Q150)
5. The translation of the 1st edition in Russian (Q7737)

My question: Should the item for the written work contain multiple title (P1476) statements for the title of the edition in English (Q1860) and the title of the translations too (i.e. Portuguese (Q5146), French (Q150) and Russian (Q7737))?

Rdrg109 (talk) 04:55, 2 April 2022 (UTC)

On the work data item, title (P1476) is for the "original" title of the work. Translations of the title into other languages should not be included on the primary data item for the work. Titles of translations should appear on the data items for those translations. --EncycloPetey (talk) 17:34, 3 April 2022 (UTC)

What about situations where you have a Latin book from say the 16th century where the printed title begins with words which generally are not considered part of the title. For example, the 1680 edition of Boethius is titled: Anicii Manlii Severini Boetii v.c. et inl. ex cons. ord. exmag. off. atque patricii Consolationis philosophiae libros quinque but the title as we know it is: De consolatione philosophiae. - Kosboot (talk) 01:20, 4 April 2022 (UTC)

For some works, there is a standard scholarly title. But you are correct, some Latin works will have lengthy titles because of past titling practices. --EncycloPetey (talk) 18:54, 4 April 2022 (UTC)

Modeling written work when there is a different number of books/edition between translation

Hi, I stumbled on a thorny case. My Brother's Husband (Q22128978) is a manga published in 4 books in Japan. In France, the split was kept identical, but while adding the french editions, I noticed that the english translation was published in 2 volumes (volume 1 being 1 and 2 of the original), which mean I incorrectly placed the edition. I wasn't able to find guidance in the project page, nor in the discussion pages, mostly because I have no idea on what to search exactly. Does anyone have a example where this is handled I could look at ? --Misc (talk) 23:52, 9 May 2022 (UTC)

Cleaning Wikidata:WikiProject Books/Participants?

Hi,

Posting the previous message, it seems Wikidata:WikiProject Books/Participants is too large to be pinged.

In itself, it's a tool problem more than a list problem. Regardless, I had a quick look and saw multiple inactive accounts. On 137 users listed, around 30 seems to have been inactive for at least a year (and I've see one user who choosed to disappear and one who only made one edit on Wikidata).

Should we clean this list and move the inactive persons to a "former participants" page? (not sure it will be enough to allow a ping but at least this list would be more up-to-date). And if so on what criteria? Inactive for at least a year (I think we have a tool for that somewhere, no?) or just some lighter clean-up (like the two edge case I mentioned above).

Since I can't ping everybody, @Sic19, Epìdosis, CaLéValab, Hsarrazin, Salgo60, Dcflyer: I know that you're the most active on this group recently, what do you think?

Cheers, VIGNERON (talk) 19:16, 2 February 2022 (UTC)

that seems a good idea...

also, since we asked for a fixing of the problem for projects exceeding 50 names, and nothing was done, because it's a feature, it's not a bug... I suggest maybe we divide the list in 50-names batches (after cleaning)... -> https://phabricator.wikimedia.org/T200350

do you think it would work ? Hsarrazin (talk) 19:23, 2 February 2022 (UTC)

I  Support the proposal, of course. --Epìdosis 20:18, 2 February 2022 (UTC)

 Done, see Wikidata:WikiProject Books/Former participants (I hope I didn't do any mistakes, I tried to be thorough). Cheers, VIGNERON (talk) 09:56, 29 May 2022 (UTC)

The Encyclopedia of Fantasy ID

My authority control proposal might interest this project. Susmuffin (talk) 12:30, 11 June 2022 (UTC)

P1922 for work or edition?

Hi,

I wrote a message on Property talk:P1922 in 2017 but I forgot about it a bit and only two people left a comment.

The general question is: where should we use first line (P1922)? For work? edition? both? only one?

My opinion is that is probably should be on edition only but I would love to have more insight on this question.

Cheers, VIGNERON (talk) 15:00, 15 May 2022 (UTC)

Hi Vigneron, if you look at the given examples "work" is the standard (Q208460#P1922) since you don't want to repeat the first line over and over again or click through the editions to find it. On the other hand you might need the property for a differing edition or differing translation. So imho P1922 is for works and editions. --Kolja21 (talk) 20:00, 15 May 2022 (UTC)

PS: single-value constraint (Q19474404) should be removed since this property can be used for different languages. --Kolja21 (talk) 20:12, 15 May 2022 (UTC)

The first line of a work can be different in different editions. For example s:en:Thanatopsis has a different first line in different editions, even though it is the same poem. --EncycloPetey (talk) 22:09, 15 May 2022 (UTC)

The first line in any language for a translation will be specific to that translation. If you are adding the first line from a French translation, that first line belongs on the data item for that translation. --EncycloPetey (talk) 22:07, 15 May 2022 (UTC)

It would be nice if you had discussed this first before deleting the beginning of the Odyssey (Q35160) in German ("that belongs on the data item for Weiher's translation"). There are hundreds of German editions of Odyssey (Q35160). Nobody wants to click through dozens of items only to find out that there is probably no first line (P1922) on the edition item. So imho it's helpful and wise to collect first line (P1922) on the work item adding the most famous version and it's translation with the name of the translator. Otherwise all the examples given in Property:P1922#P1855 are wrong. Non of the examples is an edition item. --Kolja21 (talk) 22:30, 15 May 2022 (UTC)

@EncycloPetey: Would you please stop deleting quotations! You don't even add an edition item and move the quotation. You only give orders and delete. This is unacceptable. --Kolja21 (talk) 22:41, 15 May 2022 (UTC)

It is not my responsibility to track down all the publication data for you. You have the data; I do not. If you decide to do something without community approval, it is on you to clean up the issue. --EncycloPetey (talk) 23:12, 15 May 2022 (UTC)

We do not create data items for first lines, we create them for editions. If you have information about an edition, then create that data item. I am not going to create a data item for a single line from a text without knowing any of the publication data. --EncycloPetey (talk) 23:27, 15 May 2022 (UTC)

What an arrogant answer. It fits to your misuse of the reverse button. --Kolja21 (talk) 19:22, 16 May 2022 (UTC)

@Kolja21, EncycloPetey, Kolja21: I know P1922 is currently used on a lot work items (but not mostly, only the 6th and 11th most, see https://w.wiki/5Arw which means that the given examples are not representative by the way, it need to be updated) and I'm not sure it's a good thing. An incipit is written while a work is unwritten, so it's not really logical. It's especially not good when there is several incipit with no references and/or no qualifier to distinguish them (see Q19048723#P1922 for instance). For the context, on the French Wikisource we are looking to reuse Wikidata and it's a bit problematic.

My point of view is that we should only have the original (whatever it means) incipit on the work (like we do for original title) and put the incipit of the edition on the item of the said edition (and thus we could keep single-value constraint (Q19474404) which is already respected in 96.5% of the uses). And we should probably have a reference constraint (as a suggestion, at least as a first step). What do you think?

EncycloPetey: I kind of agree with Kolja21, you should at the very least wait for this discussion to be over to remove them. That said, Kolja21, since we have no data (we don't even know if this is a real incipit), it's hard to move them anywhere if you don't know anything (which is exactly why I'm asking for references ;) ). In any case, please keep it civil.

Cheers, VIGNERON (talk) 18:48, 16 May 2022 (UTC)

@VIGNERON: on certain items, mainly old work whose text isn't entirely secure (but I imagine you could have the same thing with works that were revised by their author), we may have multiple incipit/explicit (e.g. on Q21774084 or Q111021417). I don't know if something more flexible that single-value constraint (Q19474404) exists to deal with that. --Jahl de Vautban (talk) 18:59, 16 May 2022 (UTC)

Yes, or incipits for different parts of a work, e. g. prologue and main text, which might not always be present together in all manuscripts of a given work. See also Richard Sharpe: Titulus ISBN 978-2-503-51258-7. HHill (talk) 04:16, 17 May 2022 (UTC)

We would need criteria for which "original". The first line of Thanatopsis (Q7710153) (in its first publication: Thanatopsis (Q111820194)) was added by the editor, and was not part of the poem until it was edited and printed. The author didn't write it. So, first published copy isn't necessarily a good choice. And even if we go be first publication, for novels that were serialized in a magazine before being printed as a book, do we use the first installment of the serialized edition, or the first "book" edition, or the author's manuscript? There are many, many reasons and examples where the first line (and the last line) should be placed on only a specific edition. I see that A Clockwork Orange (Q692557) has a "last line", but that line does not appear in the US edition because the US editor removed the last chapter of the book. For H. G. Wells' The Time Machine: An Invention (Q627333) the first line is different in each of the two 1895 editions (one published by Heinemann and one published by Holt). In any event, none of the current examples show the use of this property on a work for a line from a translation of the work. If we put the incipit from translations on the work data item as well, then we could have thousands of such incipit on the Odyssey (Q35160). There are potentially 100's of languages, some of which have more than 100 different published translations. --EncycloPetey (talk) 05:09, 17 May 2022 (UTC)

@Kolja21, EncycloPetey, Jahl de Vautban, HHill: so, what do we do now? Cheers, VIGNERON (talk) 08:55, 29 May 2022 (UTC)

@Kolja21, EncycloPetey, VIGNERON, HHill: well, nothing apparently I am in favor of 1) limiting the use of incipit to the language of the work; if it's a translation, it goes to the edition. 2) For constraints, single-best-value constraint (Q52060874) instead of single-value constraint (Q19474404). --Jahl de Vautban (talk) 16:16, 11 June 2022 (UTC)

@Jahl de Vautban: I'm also in favor of 1) but not sure if there is a consensus yet. For 2) I somehow missed your previous comment, I think what you want is keeping single-value constraint (Q19474404) but adding separator (P4155) in qualifier. Cheers, VIGNERON (talk) 17:01, 11 June 2022 (UTC)

Notice for has narrative theme property proposal

Yall might be interested. Lectrician1 (talk) 13:13, 1 July 2022 (UTC)

How to link a preface to the main work?

What's the recommended way to link a preface work (like Author's Preface for the Poem on the Lisbon Disaster (Q113162958)) to the main work (in this case, Poème sur le désastre de Lisbonne (Q3401061))? I found front and back matter (P8570) but I'm confused by it and not sure how it should be used. JesseW (talk) 14:36, 20 July 2022 (UTC)

This is how I've see this modeled before. --Marsupium (talk) 02:16, 27 July 2022 (UTC)

Thanks, I wasn't familiar with statement is subject of (P805). I've added some documentation to front and back matter (P8570) to hopefully help others confused like I was. JesseW (talk) 01:58, 28 July 2022 (UTC)

I wonder what the appropriate reverse property would be. It doesn't look like one is explicitly stated. JesseW (talk) 02:02, 28 July 2022 (UTC)

Most properties do not have a reverse property, and this is intentional. --EncycloPetey (talk) 04:18, 29 July 2022 (UTC)

Do you happen to know of a help page that explains more about what that intention is? What distinguishes properties where a reverse is considered useful, and ones where it isn't? JesseW (talk) 12:19, 29 July 2022 (UTC)

New property: The Encyclopedia of Fantasy ID (P10883)

A new property, The Encyclopedia of Fantasy ID (P10883), has been created today. UWashPrincipalCataloger (talk) 01:45, 25 July 2022 (UTC)

Property proposal

Hi,

I proposed a new property: Wikidata:Property proposal/CCFr document ID (for the CCFr (Q2941626)).

Cheers, VIGNERON en résidence (talk) 16:01, 10 August 2022 (UTC)

instance_of for factual books

For books that are factual, but not academic, what should they be an instance of: book (Q571), literary work (Q7725634), or written work (Q47461344)? All seem to be in use. I'm not worried about editions or individual copies, and note that simple things like page counts are deprecated for some of these options. Should they be only one thing? Vicarage (talk) 09:03, 4 September 2022 (UTC)

First, book (Q571) should never be used. There are many past community discussions that have all concluded that it is inappropriate for use. But written work (Q47461344) or scholarly work (Q55915575) can be used, depending on which is more appropriate. --EncycloPetey (talk) 14:56, 4 October 2022 (UTC)

FRBR bibliographic layer conflation in Wikidata vs. BIBFRAME

The bibliographic properties section now reads:

Not to complicate too much, we didn't use the FRBR terms "expression" or "manifestation", as the boundary between the definitions is not easy to grasp. So we collapsed those two FRBR layers in one (BIBFRAME Instance): "edition" (other conceptual frameworks similar to FRBR, like Bibframe, also collapse those two layers).

But judging from the English Wikipedia BIBFRAME article, BIBFRAME doesn't conflate the FRBR levels "expression" and "manifestation", in fact it uses its "work" entity to combine the FRBR levels "work" and "expression". Its "instance" entity is analogous to FRBR "manifestation", and so doesn't quite match Wikidata's "edition" entity (seeing that that combines FRBR "expression" and "manifestation"). Maybe the section quoted above could be changed as follows?

Not to complicate too much, we didn't use the FRBR terms "expression" or "manifestation", as the boundary between the definitions is not easy to grasp. So we collapsed those two FRBR layers in one: "edition" (in contrast, the BIBFRAME bibliographic data model instead combines the FRBR layers "work" and "expression" in a single "work" entity).

Murkwuite (talk) 06:43, 26 September 2022 (UTC)

I've now made the change proposed above. Murkwuite (talk) 07:19, 5 October 2022 (UTC)

folk tale (Q1221280) can't have an author

If you agree and there is no objections, please add this constraint, it will help to avoid confusion between folk tale (Q1221280) and fairy tale (Q699). Probably instances of folk tale (Q1221280) should have a mandatory 'novalue' in P50, or they shouldn't have this property at all. Podbrushkin (talk) 12:41, 13 October 2022 (UTC)

@Podbrushkin transcriber (Q54855620) seems more appropriate to me. Jarnsax (talk) 18:27, 15 October 2022 (UTC)

While a folk tale (Q1221280) itself may not have an author, editions of folk tales may have an author, who told that particular tale in a particular way. --EncycloPetey (talk) 02:41, 16 October 2022 (UTC)

@EncycloPetey: no and yes. I would prefer to see an editor for the work in which it appears rather than encourage recording an author of folktale, unless they have some particular spin on it that makes it notable. I put up some of the Australian indigenous dreaming stories where they were stated as being "related to" the stated author of the work, and didn't feel comfortable recording them that way and still haven't done the subparts to WD due to that. s:More Australian Legendary Tales  — billinghurst sDrewth 04:57, 16 October 2022 (UTC)

collection of short stories

Hello, I think short story collection (Q1279564) should be a subclass of literary work (Q7725634), and not of written work (Q47461344), because every short story (Q49084) - is a literary work. Also look at this tree, stated in project page: https://w.wiki/5PMd, short story collection looks out of place in here. Podbrushkin (talk) 21:20, 7 October 2022 (UTC)

Hmmm. I can see both ways. I would not necessarily agree as "work" implies a unity of a single object, and a "collection" is usually not a unity but rather a collection of things. --EncycloPetey (talk) 02:41, 8 October 2022 (UTC)

Do you mean it shouldn't be a child of neither of them? Maybe it's true, but I've proposed a minor change which wouldn't affect existing hierarchy or sparql queries much, but will make a tiny improvement. Podbrushkin (talk) 16:11, 10 October 2022 (UTC)

I think a 'collection of short stories', by various authors, is actually an instance of compilation (Q1614239).. it's a copyrightable "work" by the editors, that includes "works" by other people.

The same would, in a more general sense, apply to when an author is editing a collection of their own writings... the collection is a "new worK" that is compiling separately created and 'self-contained' "works" (previously published or unpublished really doesn't matter).

Not to argue that such things are not also a "literary form". Jarnsax (talk) 21:30, 16 October 2022 (UTC)

Apart from compilation (Q1614239) there are: series of creative works (Q7725310) group of works (Q17489659) collection of literary works (Q108329152) group of literary works (Q17518870) literary cycle (Q867335) fiction series (Q15980953) book series (Q277759) Podbrushkin (talk) 17:32, 17 October 2022 (UTC)

I think those are more 'abstract' ideas... they don't specifically refer to a group of works of a specific 'type' (short stories) or imply they were issued as a single physical publication like short story collection (Q1279564) does. Jarnsax (talk) 20:31, 18 October 2022 (UTC)

Object to removing identifiers

@Maxlath: I object to the removal of identifiers per https://www.wikidata.org/wiki/Wikidata_talk:WikiProject_Books/2020#Clarifying_the_use_of_identifiers_for_editions_of_written_works_on_written_works_themselves.: If you are going to remove an ISBN number, I think you should make sure that there is a separate Wikidata item for the first edition in addition to the concept of the work without an edition. That's extra work, which I find objectionable.

• If I recall correctly, a Wikidata item I created was deleted by another editor, because it did NOT have an ISBN number or some similar identifier.
• Even if the original item is retained, its value is reduced, because of the loss of the identifiers that were deleted.

If you are going to delete identifying information, I think you should first copy the item, e.g., from "A" to "A1", then make "A" the item for the general concept so "A" has edition "A1" and "A1" is an edition of "A". Then you can delete the identifying information from "A", because it's retained in "A1".

However, that's extra work that I do not think should be imposed upon a typical editor. I like the fact that there is a bot that converts an ISBN-13 without dashes into a proper ISBN-13 with dashes in the positions where Wikidata thinks they belong. I enter the numbers for and ISBN-13. Wikidata complains that I didn't do it right, and the bot fixes the problem not too long thereafter.

Only yesterday, I found at least one book that was a previous edition, but I wanted to cite a different edition. So I fixed it so the previous edition "had edition or translation" being the one I just created, and my newly created item was an "edition or translation of" the earlier one.

Creating Wikidata items already seems to be more work than using the book template in Wikipedia, unless the item is going to be used more than once. I routinely create Wikidata items, because (a) they tend to be more complete and consistent, and they are easier to (b) resuse and (c) maintain.

PLEASE do not make it harder to create Wikidata items.

Thanks, DavidMCEddy (talk) 13:56, 17 October 2022 (UTC)

@DavidMCEddy: I hear what you say, but letting edition identifiers on works contribute to blur the distinction between work items and edition items. This blurriness and the tendency for items to switch from work to edition roles makes Wikidata painful to use as a bibliographic database. I'm not so familiar with the use in Wikipedia templates: is it that the template expects both work and edition data on the same item, or can it work with two distinct items? Could you give me an example of an article using such a template, and where removing the ISBN would be an issue? -- Maxlath (talk) 14:27, 17 October 2022 (UTC)

@Maxlath: Thanks for the reply and question. See Wikipedia:Wikipedia:Citation templates, Wikipedia:Template:Citation, and similar pages, including in other languages. As far as I know, those are still the standard, recommended way to cite published works.

However, each citation to a particular document must be completed from scratch in the new location (or copied from another use, if a user can find that). This creates duplication of work and is difficult to maintain, e.g., with link rot.

Some Wikimedia Foundation projects allow something like Wikipedia:Template:Cite Q as an alternative. This is less work for citations used in multiple places and makes maintenance much easier. However, since that's cryptic in source view, I routinely include a comment with such references, e.g., <!-- Author (2021) Name of document -->{{cite Q|Q12345}}.

Does this answer your question? Thanks, DavidMCEddy (talk) 15:53, 17 October 2022 (UTC)

@User:Maxlath @DavidMCEddy Just to point out how I'm doing it... see w:User:Jarnsax/citations/publishers (these books, and their bits, are cited various places on en, that I have been fixing) and the linked entities.

I'm storing the "multiple" identifiers that refer to different formats on the entity, and just qualifying them as whatever 'distribution format'. I think this makes more sense, from a 'bibliographic' standpoint.... all 'versions' that have the same text block are "imprinted from substantially the same setting of type", to use the old definition, and so are the 'same edition', just different issues or reissues. Particular bindings are basically just classes of exemplars.

I'm also using that markup in articles 'as written' on that page, with additions in the article if it needs some other style of display.... to be specific, the "author name stuff" is needed since Cite Q doesn't have finished code for that yet, and sometimes you need to hide it for a specific citation. Still having it 'there' but hidden should let the template create the Coins metadata for the cite (I hope, lol).

BTW, the citation templates do not work if a source has an editor and contributors, but no actual author... they don't display and you get a hidden error, so use "others" and write it out.

The explicit hidden comment (which I think is a very good idea) will hopefully make editors less likely to claim that Qids are 'obscure'. You commented about doing the same sort of thing... if the generated CITEREF tag is going to be obscure from the markup (which can completely omit the author's name, year, etc) just add a comment. People understandably want to be able to know what the source is (and how to repeatedly cite it) from the markup.

Just as a aside, you can easily check/convert/hyphenate many ISBNs at https://www.loc.gov/publish/pcn/isbncnvt_pcn.html instead of relying on a bot. It knows about most publisher prefixes (all English ones, I believe). For the ones it doesn't, if you put the ISBN into the "Global Register of Publishers" (at https://grp.isbn-international.org/ ...a free account is highly recommended, it's somewhat useless otherwise) it will show you the matching ISBN prefixes (which tells you how to hyphenate it). The LoC tool will still convert from 13 digit to 10 digit and back, and indicate bad check digits, for ISBNS it can't hyphenate.

One common issue when looking at enwiki cites is someone 'converted' from 10 digit to 13 digit by just adding 978- to the front, without recalculating the check digit. The LoC tool makes this obvious.

I think the main thing we should keep in mind is that these identifiers are not really properties of the "subject of the entity", even if it's an ISBN printed in the book. They are merely pointers to external (proprietary, usually) databases.

Conflating "work" identifiers (there are not many) with "edition" identifiers is just going to confuse the bots trying to update things like Worldcat. It's even worse with UPCs (which is what ISBNs are, they are for wholesale distributors and POS terminals, not us) since a UPC can 'refer to' something without actually being printed on it.

That's the source of a lot of the "junk" in Open Library... imports of book data from distributors, that includes 'linked' ISBNs (such as a display stand for a new book, that comes with enough copies to fill it, that is ordered as a single item, or separate books sold shrinkwrapped together). Linked ISBNs are useless to anyone but a bookseller.... they are egregiously ephemeral, and sites like Amazon quietly forward you to the right item.

Please don't create a new entity just to carry different ISBNs. Being cited somewhere doesn't make an 'exemplar' notable.

I think its 'okay' to just create the edition entity for ease of editing, since it implies that such a work exists (and someone else can add it), but actually storing "work" identifiers onto "editions" (even 'until fixed' unless done relatively quickly) is likely going to break re-users of our data.

In case it's not obvious, I'm working through the UC Press backlist as a source of 'examples' since they are widely cited in lots of various use cases (often badly) to hunt bugs/omissions, and try to establish some idea of 'best practice'... I'm being very particular about the displayed cites being of exactly what the editor originally said, and "bibliographically complete". They also have a lot of open access books... my spreadsheet is much more extensive and not 'imported' yet, lol. I'm doing this stuff manually, by deliberate intent.

The one caveat to citing 'exactly the same thing'.... it's not uncommon that Google Books allows preview of the "best version" only... so you can sometimes look at and 're-cite' the same statement from there, to a actual print edition, with a page number and link (this does not work for ebook editions with reflowable text...no fixed page numbers). Only do that if you've actually looked at the page/snippet... adding search terms to the cite (see w:Template:GBurl) is effective if the particular page isn't in the preview. Colophons are also wonderful to look at, and almost always in the preview, even if at the back of the book.

I've located transposed page numbers in cites this way. Jarnsax (talk) 22:04, 18 October 2022 (UTC)

Thanks, DavidMCEddy (talk) 00:21, 19 October 2022 (UTC)

@DavidMCEddy I know that was kinda a rant and infodump, lol. Jarnsax (talk) 00:32, 19 October 2022 (UTC)

Anthologies, "published in", "part of", works and editions

I have run into a disagreement with @EncycloPetey that we have been unable to resolve between us, but that seems to be of wider interest, so I would be interested in feedback. The discussion is at [1], and to me it seems there are at least three areas of disagreement: 1. Whether anthologies of short stories are "literary works" 2. Whether "has part" is the correct property to use as the inverse of "published in" for anthologies 3. Whether "published in" should be used at all to link literary works to anthologies they are published in (rather than only for editions) Personally, I remain convinced that the current data stucture and documentation of Wikidata points to "yes" as the answer to all three questions, but @EncycloPetey seems to think this not just in some general sense incorrect but (at least in part) counter to "actual usage". Unfortunately, I have not been able to find sufficient examples of well-modelled literary anthologies to settle this from established usage. I would be interested in additional perspectives on this. Pfadintegral (talk) 12:59, 24 October 2022 (UTC)

Property proposal : USTC identifier

I have just made a property proposal about USTC identifiers. If you are interested in early modern books (1450-1650) you may have heard of this database of more than 800000 editions, with information about more than 4 million hard copies and links to more than 200000 digital copies, available at https://www.ustc.ac.uk/. FreeCorp (talk) 08:02, 3 November 2022 (UTC)

Help page for works and editions

I recall a help page somewhere on Wikidata describing the distinction between works and editions, but I can't seem to find it at this time. Any ideas? Daask (talk) 19:48, 16 November 2022 (UTC)

@Daask: Not on wikidata.org, but we did one on wiki.inventaire.io that largely overlaps with Wikidata Books considerations https://wiki.inventaire.io/wiki/Guides#Editions, could that be it? -- Maxlath (talk) 11:22, 18 November 2022 (UTC)

Property proposal : ISFDB award id

Wikidata:Property proposal/ISFDB award ID adds to the 5 other Internet Speculative Fiction Database ids we have. Please support it Vicarage (talk) 18:46, 24 November 2022 (UTC)

Wikidata property related to editons

We have Wikidata property related to creative works (Q18618644): Wikidata property for items about (creative) works and objects (including buildings) with facet of (P1269): work (Q386724) and version, edition or translation (Q3331189). Shouldn't works and editions keept separate? We could create "Wikidata property related to editons" for properties like Library of Congress Control Number (LCCN) (bibliographic) (P1144) and K10plus PPN ID (P6721). --Kolja21 (talk) 08:39, 13 December 2022 (UTC)

The work / edition distinction is used in bibliographic databases. It doesn't apply across all kinds of objects for which the item is used, so the naming and distinction cannot be made specific for bibliographic items. --EncycloPetey (talk) 00:52, 15 December 2022 (UTC)

Ping @UWashPrincipalCataloger: who aded facet of (P1269) = version, edition or translation (Q3331189) (Special:Diff/1471913754). Indeed this seem quite stange (here "works" is not the FRBR level, it's any work, bibliographical or otherwise). Cheers, VIGNERON (talk) 09:44, 18 December 2022 (UTC)

Some data models (e.g., BIBFRAME and WorldCat Entities) lump FRBR works and expressions together. In those models, a translation is treated as a work rather than an expression of a work. So in some situations this property is indeed a facet of version, edition or translation (Q3331189). UWashPrincipalCataloger (talk) 04:24, 19 December 2022 (UTC)

@UWashPrincipalCataloger: sure, but what does it have to do with Wikidata property related to creative works (Q18618644) which is *not* specific to books (nor work nor edition).

I guess I am thinking that it doesn't always have to be a facet of the value (such as in the case of a building, but for bibliographic resources, it can be. Is that statement incorrect if it only applies to some kinds of things? UWashPrincipalCataloger (talk) 20:32, 19 December 2022 (UTC)