Wikidata talk:WikiProject Books/2016

Welsh books

Hello, User:Ivan A. Krestinin informed me that my bot made some mistakes. The bot run was requested at the Bot requests page and I ran the bot as described at Wikidata:Requests for permissions/Bot/FischBot 5. Wikidata:Database reports/Constraint violations/P1104 shows a long list of constraint violations caused by my bot run. What should I do about these constraint violations? --Pyfisch (talk) 19:31, 11 January 2016 (UTC)

Can we use author (P50) for author of preface or foreword ?

I just want a confirmation that we can use author (P50) to specify the author of preface or foreword and by adding the qualifier applies to part (P518): preface (Q670787). If yes can we specify that on the main page of the project to be sure that will use by everyone ? Thank you Snipre (talk) 14:28, 13 January 2016 (UTC)

Can you give an example? I just know the case contributor to the creative work or subject (P767) with the qualifier applies to part (P518): preface (Q670787). --Kolja21 (talk) 21:05, 13 January 2016 (UTC)

Kolja21 See From the Pharaohs to the Fall of the Roman Empire (Q22003784). Snipre (talk) 21:25, 13 January 2016 (UTC)

A book with 5 authors looks strange. I would expect one or two main authors and the rest of the people have special functions like editor, contributor etc. You can make a test with {{Cite item}}:

• Jean Leclant, Jean Vercoutter, Frank M. Snowden, Jr. and Jehan Desanges, Ladislas Bugner (eds.), L'image du Noir dans l'art occidental, Paris: Éditions Gallimard, 31 January 1991, ISBN 978-2-07-011197-8

--Kolja21 (talk) 21:48, 13 January 2016 (UTC)

Kolja21 This is a collective work so there are several authors. Snipre (talk) 12:42, 14 January 2016 (UTC)

If they all wrote all of the chapters, then they are co-authors. If they each wrote separate sub-parts they are contributors. Generally when that is the case there will be an editor assigned.  — billinghurst sDrewth 13:19, 14 January 2016 (UTC)

Again, it is a collective work: no particular role for each author. No editor. The only one who is different is the author of the preface who is recognized as only auhtor of this part. But please consider the question of the preface only. We know that the preface was done by only one person. How can we specify that characteristic ? By using author (P50) with qualifier or should we create a new property ? Thank you. Snipre (talk) 15:24, 14 January 2016 (UTC)

See above: For cases like this we have contributor to the creative work or subject (P767) with the qualifier applies to part (P518): preface (Q670787). But if you want a new property "foreword by" it's ok with me. --Kolja21 (talk) 19:21, 15 January 2016 (UTC)

Narrative location used in map

I have used the narrative location (P840) for novels, short stories, poems, non-fiction books to make a map of narrative locations in Denmark (Q35). You can see an example here: http://fnielsen.github.io/littar/ This application has recently won a second prize in an app competition organized by Danish Bibliographic Centre (Q12307383). genre (P136) is used to color the placemarks and the quotation (P1683) qualifier provides text for the placemarks.

Here are some methods I apply:

1. Rather than using broad locations such as Denmark (Q35) or Copenhagen (Q1748) I tend you use local location such as streets, buildings, café whenever possible.
2. To source the statement I use the quotation (P1683) qualifier with a snippet from the literary work to make it apparent that the work actually has the narrative location. For copyright protected work I select a single or two sentences to be within fair use, while for copyright-expired works I sometimes extract a bit longer text (at most 400 characters).

So far I have mostly done it for locations in Denmark, but the scheme can of course be applied for other countries. I note that Patrick Modiano (Q222944) tend to be very good with locations. I have added a few locations to Missing Person (Q3452094) from the first part of the book. I have not added the quotation (P1683) qualifier, but I suppose you can grasp the idea. You can see a Danish work with location labeling here The Fall of the King (Q1758876).

— Finn Årup Nielsen (fnielsen) (talk) 01:11, 21 January 2016 (UTC)

@Fnielsen: that a fantastic application, could you do it for France (Q142) too ? I had a similar idea for books about Rennes (Q647) (I already take care of s:fr:Portail:Rennes) but no idea on how to do it ; thanks for giving me the structure. Cdlt, VIGNERON (talk) 08:08, 1 April 2016 (UTC)

section, verse, paragraph, or clause (P958) for song numbering

I have begun to use section, verse, paragraph, or clause (P958) for song numbering in song book (Q1650727). I tried to start a discussion here: Property talk:P958, but noone has reacted. I have previously suggested an entire new property for the case, see item number/enumerator. series ordinal (P1545) could also be used. Is this better? — Finn Årup Nielsen (fnielsen) (talk) 20:43, 22 January 2016 (UTC)

Maintenance to be done.

Hi. I have just realised that I have been labelling editions of books with original language of film or TV show (P364) rather than language of work or name (P407). Would someone like to fix those up with the available tools? If there is not already a violation checker in place for that error, it might be worth putting one in place. Thanks for whatever you can do, or suggestions that you can make.  — billinghurst sDrewth 04:05, 2 February 2016 (UTC)

@Billinghurst: At http://tools.wmflabs.org/autolist/index.php?language=en&project=wikipedia&category=&depth=12&wdq=claim[364]%20AND%20claim[629]&pagepile=&statementlist=&run=Run&mode_manual=or&mode_cat=or&mode_wdq=not&mode_find=or&chunk_size=10000 there is a list: you can use "claim[364:NUMBER OF THE ITEM OF THE LANGUAGE] AND claim[629]" in "WDQ", then in the lateral box use

-P364:NUMBER OF THE ITEM OF THE LANGUAGE
P407:NUMBER OF THE ITEM OF THE LANGUAGE

and "Process commands". Repete this action for all the languages. I don't know if there are better solutions. --Epìdosis 06:26, 2 February 2016 (UTC)

K, thanks. Looks ugly. :-/  — billinghurst sDrewth 11:18, 2 February 2016 (UTC)

Classification of essay (Q35760)

I think the current classification of essay (Q35760) is incorrect. As you can see below, it is correctly marked as a subclass of work (Q386724). However, it is also a subclass of literature (Q8242), which is a subclass of occurrence (Q1190554). In my opinion, saying that an essay is an event is wrong. Compare with short story (Q49084):

A short story (Q49084) is a work (correct) and a text (correct), not an event. So how do we fix essay (Q35760)? Mushroom (talk) 16:33, 29 February 2016 (UTC)

The problem is literature (Q8242) for essay compared to literary work (Q7725634) for short story. IMHO we should distinguish between literature and literary work and connect essay to literary work. The main problem is the definition of literature (Q8242): currently this is a subclass of art and a subclass of text. This means we should consider literature only as an art and not as a text. But I am not so sure about that. Snipre (talk) 18:17, 29 February 2016 (UTC)

I think that essay (Q35760) should be subclass of literary work (Q7725634), which is "part of: literature".--Micru (talk) 21:33, 29 February 2016 (UTC)

Review of "works" and "editions"

Hi everyone. After months (or years) of absence, I'm taking the courage to come back here and seek to solve my confusion about how to deal with books in Wikidata. I've just returned from a conference, meta:WikiCite_2016, where we discussed at length about this with many librarians, scientists, Wikidata experts. My immediate goal is to solve the issue of the integration between Wikisource and Wikidata, although it seems that some Wikisource community already started the job. I'd love to know how they are doing that, and if there are any issues.

During the conference, we quickly reviewed properties relative to "works" and "editions", and I'm kinda happy there were no major flaws, apparently: still, some tweaking is probably due.

For example, and I'm just starting the discussion, we thought that putting "instance of book" for items at the "work level" is not really right: when we talk about Pinocchio, Hamlet, Pride and Prejudice, we have WD items linked to Wikipedia articles, which talk about the work in general, the work that has inspired books, but also movies, cartoons, musical. So we should define another way. "instance of book" could be used only for "edition-level" items.

I'm exploring with OCLC folks some identifiers they use for works and editions, so I'll be back with more news: apparently, there was some confusion in our use of OCLC control number (P243) (which should be used at "edition-level" and not "work-level".)

Other thoughts:

• "edition level" items should not have links between each other, but when they are direct translations. The "work level" item of course should be connected with them all.
• properties like author, title, original language of work can be both in the work and the edition.
• we are not "cataloging the universe": the first goal is to connect Wikimedia projects. Wikipedia, Wikisource, Wikiquote.
  • in this sense, "work level" items are created for Wikipedia articles, and when we have multiple editions on Wikisource.

Some other notes will follow. Aubrey (talk) 23:39, 29 May 2016 (UTC)

Welcome back Aubrey ;)

About OCLC control number (P243), Hazmat2 left a message about the same subject on Property talk:P243.

There is something strange about using book (Q571) for the first level of FRBF, very strange as : the item book (Q571) is about a physical object (made of paper !) and the work-level of FRBR is about a conceptual idea. If I understand correctly your remark about Pinocchio, you are suggesting to widen the item for the first level. If it is it, I agree completely ! FRBR come from the librarian but can be applied for all documents not just book and we should use something more like work (Q386724) (or probably a new item made for our need).

« No links between edition » seems über-obvious, I don't remember seeing that.

« properties like author, title, original language of work can be both in the work and the edition. » I'm not a big fan of redundancy... I can understand the need to replicate the author (useful for identifying more easily both the work and the edition) and the title (not redundant since the title change nearly always for each edition) but why the original language of work as the name say it only about the work ;)

I find that « not "cataloging the universe" » is very wise ; we'll maybe do that at the end of infinity but clearly not now while the structure need some big improvements.

Cdlt, VIGNERON (talk) 06:37, 30 May 2016 (UTC)

It's great to hear that this stuff is being thought about. I've been tinkering with a little tool for browsing Wikisource books (and their proofreading status, epub downloads, etc.) and have been basing it where I can on the data in Wikidata. In the process, I've been trying to get my head around how things are done here and how they're "supposed" to be done here and what the plans are for the future. It actually all makes quite a lot of sense! Which is nice. :-) Some thoughts:

• The repetition of properties on edition-level items is only done where these differ from the work-level values.
• version, edition or translation (Q3331189) is a subclass of book (Q571) which I think is not correct (but I can't tell you what it is a subclass of). There are currently 63 subclasses of book. Perhaps this is just me rebelling against substitutability though.
• There's no need for a publication date (P577) on the work-level item, because there should be an edition-level item with this property and it should be the earliest one. This is just one example of redundancy that perhaps should really be allowed to remain because it does make it easier to query things. Is that a permitted de-normalization?

Anyway, it's great fun working through the problems. I'm just aiming at a simple little interface that presents what's in (any language) Wikisource along with links to download, proofread, etc.

Samwilson (talk) 07:24, 30 May 2016 (UTC)

Thanks guys for weighing in. These things are complex enough doing alone, so it's very good to check with others :-D So:

• OCLC number: it seems to me this is really at the edition-level. There is another OCLC id for the work, I'm waiting for instructions.
• I was not convinced by work (Q386724) for our higher level, because I thought that people in Wikidata use is in a different manner, but after reading the its description I think is abstract enough. What I do not understand is how to specify, maybe, that in the origin, the first instance of the work had been a book...
• I think some redundancy is needed for the sake of simplicity. @Tobias1984: had some compelling argument that with the right query is very good to have at least author and title as redundant. As maybe publication date (P577).
• @Samwilson: Show us the tool! I'm excited. Aubrey (talk) 09:19, 30 May 2016 (UTC)

• Here it is (not that there's much to see yet): https://tools.wmflabs.org/ws-search/ Samwilson (talk) 16:07, 11 June 2016 (UTC)

• Hi all! I also just returned from WikiCite with my head full of new ideas. User:Aubrey made an amazing effort to explain the details of book cataloguing to everyone at the conference and we decided on some realistic short term goals for Wikidata. My first priority would be gathering more work- and edition-identifiers and setting up constraints and queries so we can patrol and clean our data more easily. Today I went looking for the OCLD-Work-ID (https://www.oclc.org/developer/develop/linked-data/worldcat-entities/worldcat-work-entity.en.html), but couldn't find it. Does it have a different name on Wikidata? If we don't have it we should propose it. --Tobias1984 (talk) 18:22, 30 May 2016 (UTC)
  • From what I've seen, there is no Wikidata property for the OCLC work id. The problem is that I found it but there were some issue (many URLs were not working on OCLC) so I asked Merilee (who was at the conference and works for them) to check on it. What we can also check is OpenLibrary, which has one items for works and one for editions. @Kcoyle: is also one of the leading expert of FRBR in the world so she can tell us if this simplified model is ok. Aubrey (talk) 21:00, 30 May 2016 (UTC)
  • Also, a question: is there a possibility to shift a number of statements from an item to another? I mean literally copy and paste and cut and paste. That would come very handy when we have to create the edition item of a work, or viceversa, or both. Think about when you have citations from one edition level item A to another, and then you need to create a work level for item A, A'. In my view, citations (cites work (P2860)) should shift from A to A'... So it's both a technical, practical problem and procedural one. Aubrey (talk) 14:28, 1 June 2016 (UTC)

• A fellow WikiCite participant alerted me to the notion of superwork (Q24261960). The encyclopedic articles tend to describe superworks. I suppose. — Finn Årup Nielsen (fnielsen) (talk) 12:33, 6 June 2016 (UTC)
  • Never heard of it... But the more I dig into Wikidata and books, the more I'm convinced that FRBR framework must be taken with a grain of salt: of course, the concept of "work" and "edition" is useful enough, and it maks sense to use it here on Wikidata. But, IMHO, it makes sense because for practical reasons: sometimes we have an item for a single book, sometime for the original, seminal work. We'll have translations so we need the original. But we, I think, should resiste the urge of "cataloging" the world, Giving Order to the Universe. Because if we try to do it we fail. For example, Pinocchio (the Disney cartoon) it's a derivative work of Pinocchio (the tale), but also a work on it's own, which has been traslated and adapted, ecc. So, I believe that if we manage to use "work" and "creative work", here on WD, as a simple statement, without the full FRBR connotation of "the Idea of the Work", we can do better. A. lice in Wonderland started a book, so the work-item can just say it's that. It had an original language, an original date of publication. Pinocchio started as a "feuilleton", and then it had many editions and translations... And so on. If we try to convey this basic meaning (book A is derived from book B), and start from Wikipedia articles about books and Wikisource, I think we can manage to not make too much noise :-) Aubrey (talk) 10:50, 12 June 2016 (UTC)

Claims for items of work and of volume, P361 and P1433

Hi, has a discussion and question. There are item for the multivolume encyclopedia (main work) and items for its each volume. In article items are claims published in (P1433) and part of (P361). Question: which the claims with which the items to connect? published in (P1433) the encyclopedia and part of (P361) the volume, or part of (P361) the encyclopedia and published in (P1433) the volume?

I think Wikipedia templates and other tools should immediately get from article items the name of encyclopedia and the link to it. Therefore, it should be specified in the statements of the article item.
Else if only the volume (and its reprints) items will be listed there, then to get the main item of encyclopedia (at least get the name of work) the tool will have to sort through list of volumes items, load each for search name of main work. This unnecessary problems for programmers and increase the burden on WD at least 2-3 times.

I didn't find the definition for it in the rules. According Wikidata:WikiProject_Books#Qualifiers volume (P478) is qualifier for P1433, so set a item of volume in P1433 is incorrect? Also I find that bot owners does set in P1433 and P361 duplicate of the main work item, but volume items don't set at all. E.g.: EB-11 / Descartes, René (Q20635302), EB-9 / Hans Baldung Grün (Q20205006), René Descartes (Q20635301), in ~16 thousand items of "Otto" like Q23875310. --Vladis13 (talk) 22:40, 5 June 2016 (UTC)

Queries

I started a page here Wikidata:WikiProject Books/Queries that can gather interesting and useful queries about Books. Please add your examples! --Tobias1984 (talk) 18:34, 10 June 2016 (UTC)

That's a terrific idea. :) I'll add some snippets I've been using. Samwilson (talk) 10:01, 11 June 2016 (UTC)

Great idea Tobias1984 !

Your query give me an other idea fo query :

SELECT distinct ?item ?itemLabel ?article WHERE {
	?article schema:about ?item ; schema:isPartOf <https://fr.wikisource.org/>.
	?item wdt:P31 wd:Q571 .
	SERVICE wikibase:label { bd:serviceParam wikibase:language "fr" }
}

Try it!

WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. : can someone confirm me that most this items are wrong? (the query maybe must be more refined) Wikisources contains version, edition or translation (Q3331189) and not book (Q571) and either the instance of (P31) must be changed, either the sitelink to Wikisource must be removed (and a new edition item must create).

Cdlt, VIGNERON (talk) 14:28, 13 June 2016 (UTC)

VIGNERON: As I said on Wikisource, yes, these items are wrong, but changing them will break interwiki links. So before breaking anything, we need the possibility to create interwiki links from editions. So in short, at this stage, we shouldn't create editions, unless we already have several editions of the same work (in this case, we need a "editions" page, which can get the interwikis, i.e. Q24574622 I just created). Regards, Yann (talk) 14:48, 13 June 2016 (UTC)

Ooups, I didn't see your answer ont the french wikisource scriptorium, sorry Yann.

You're right, but only when there is more than one wikisource sitelink, when the wikisources pages is not a disambig page about a specific work and when the sitelink is not hardcoded the old way on the wikisources (is it possible to do a query for that? at least for the first part?). So, there is a some items right (La Fontaine's Fables (Q2748292) ; but it's not an exception to my previous query as it doesn't have instance of (P31) = book (Q571)) and more important, there is a lot of items that can be corrected without affecting the wikisources. Cdlt, VIGNERON (talk) 15:03, 13 June 2016 (UTC)

Obviously when there is only one page in the whole Wikimedia for an item, no problem. ;o) I am not surprised about these types of errors, as the software asked for a date of publication for a work. There should only be a date of publication for an edition. Regards, Yann (talk) 16:02, 13 June 2016 (UTC)

Obviously ;) In French to be sure to understand : à quelle moment le logiciel demande-t-il la date de publication pour une œuvre ? La page Wikidata:WikiProject Books dit bien que c'est à utiliser sur les éditions (après, tout les pages se sont pas forcément à jour, il suffit de les corriger). Cdlt, VIGNERON (talk) 16:54, 13 June 2016 (UTC)

In the list of confusing things, I just noticed that I was wrongly using Q397239 instead of Q3331189. The description of these should be fixed. Regards, Yann (talk) 16:51, 13 June 2016 (UTC)

This can easily be fixed, I'll do it right now. Cdlt, VIGNERON (talk) 16:55, 13 June 2016 (UTC)

 Done There was only 5 item with instance of (P31) = editing (Q397239) and linked to Wikisource, all corrected (I'll checked other items). Cdlt, VIGNERON (talk) 17:00, 13 June 2016 (UTC)

Errors in interdescriptions of Project

Could someone do correction localization marks in description of Project? Or give link to HowToDo. Now interdescription is bad: many paragraphs have no access to translate, some titles have mistake language of translation. --Vladis13 (talk) 23:03, 11 June 2016 (UTC)

Authority data from Czech National Technical Library

Hi all, we have received an offer to import a bunch of authority data from the Czech National Technical Library into Wikidata. They offer the data in Dublin Core or (richer) MARC formats, have you heard of a way to import these directly? Or how would you approach such task and which format (out of those two) would you prefer for that? Thanks --Vojtěch Dostál (talk) 14:02, 16 June 2016 (UTC)

@Vojtěch Dostál: Please have a look at Wikidata:Data_donation. Snipre (talk) 14:34, 16 June 2016 (UTC)

@Snipre: thanks for the link. Is there something relevant to my question? Which of the mentioned tools is the one I am looking for? Thanks for clarification. --Vojtěch Dostál (talk) 15:03, 16 June 2016 (UTC)

@Vojtěch Dostál: Read Wikidata:Data_donation#How_to_add_data_to_Wikidata and then Wikidata:Data_donation#Online_tools or Wikidata:Data_donation#Bots. Snipre (talk) 18:11, 16 June 2016 (UTC)

Again, there is nothing about the file formats that I am asking about. Let's wait if other people following this discussion can help. Thanks for your efforts. --Vojtěch Dostál (talk) 18:55, 16 June 2016 (UTC)

Vojtěch Dostál I think you don't understand the process: there is no good or bad format. The good format is the format which can be read by the person who will do the import. Good format can be txt file or spreadsheet like excel if this is the format which can be handled by the system which will do the data import. A lot of bot operators are using python. The good format is not function of Wikidata because there is no unique system to import data. If you are not able to code the program which will do the import, you need to find a person who will do it and this person will say if Dublin Core or (richer) MARC formats are appropriate for their system.

So if you read the page I propose you the good answer should be "where can I find a bot operator" or "how can I use Mix'n'match/QuickStatements tool". But if you prefer to wait for someone who can say which format is the best go there. Snipre (talk) 19:45, 16 June 2016 (UTC)

I think you do not understand me. We have bot operators in the Czech Wikidata community. Also, it is fairly easy to use tools such as QuickStatements which obviate the need for bots. I am asking for advice from people who have already done authority data import and may suggest me a convenient workflow. --Vojtěch Dostál (talk) 19:55, 16 June 2016 (UTC)

Ok, I don't understand what you need. Snipre (talk) 15:38, 17 June 2016 (UTC)

verse/prose - alexandrin/octosyllabes, ...

Hello !

I'm looking to the way to indicate if a book is in verse or in prose (and we could be more specifics : free verse, poetic prose, calligram, typographic poem,...)

How to do that ?

In the same idea, how to indicate the size of the verse (alexandrins, octosyllabes,...) and the size of the versets (quatrains, tercets,...) but also the type of rimes (croisées, plates, embrassées,...) ?

Thanks for your help! --M0tty (talk) 15:06, 22 June 2016 (UTC)

@M0tty: For the size of the verse it was approved used metre (P2551); for the others you have to propose new properties at Wikidata:Property proposal. --Epìdosis 15:21, 22 June 2016 (UTC)

Book / work

Hi, I was tired to see the item whose description is "object made of paper..." for written or art books works, so I just created Произведение (Q23817623).

Imho any book-edition should be an edition of an instance of this item. author  TomT0m / talk page 20:27, 12 July 2016 (UTC)

I strongly agree!!! Do we need a bot to substitute massively instance of (P31)book (Q571) with instance of (P31)Произведение (Q23817623)? But firstly I ask you @TomT0m:: which is the difference between literary work (Q7725634) and Произведение (Q23817623), why the first would be more appropriated then the second for a literary work? --Epìdosis 21:01, 12 July 2016 (UTC)

@Epìdosis: I was thinking that not all books are textual, I'm pretty sure we can find workmade of only pictures ... I made it a subclass of "text" but I think it's incorrect. author  TomT0m / talk page 14:06, 13 July 2016 (UTC)

I agree that it is too common term, offtopic the "project books". Because work can be "film" or "work on factory". Definition should be in title, not in subclass which no one is looking. --Vladis13 (talk) 15:14, 13 July 2016 (UTC)

If i not mistake, a bunch of pages with only pictures is also "book"? Also is comics (Q1004), cartoon (Q627603). --Vladis13 (talk) 15:20, 13 July 2016 (UTC)

Lost and fragmentary literary works

We have a lot of lost literary works items, but most of them have simply instance of (P31)book (Q571). We have the item lost literary work (Q1585442), but it isn't clear how to use it (an item must have instance of (P31)lost literary work (Q1585442) only or instance of (P31)lost literary work (Q1585442) + instance of (P31)book (Q571)?). Finally, we don't have any item for fragmentary literary works. So, do we need a property like "book conservation status" or we have to use instance of (P31) (and in which way do we have to use instance of (P31)? Thank you! --Epìdosis 11:04, 25 July 2016 (UTC)

I would have though that what you are mentioning are separated instances of the same item. A book is a book (where it is lost of found); and a lost literary work is a lost literary work, whether it is published or not, let alone the format of the publishing. Re fragments, somewhat similar, the state of current copies or known copies should be separated from the condition, or how it is found. A building is a building, and at some point it remains, it is a ruin or it is removed, still makes it a building.  — billinghurst sDrewth 23:48, 26 July 2016 (UTC)

@Epìdosis: I strongly advice you to contact @Aubrey:, which will in turn know who to direct you to. --Sannita - not just another it.wiki sysop 22:10, 8 August 2016 (UTC)

I don't really know. I am having a similar issue with instance of (P31)book (Q571) and instance of (P31)literary work (Q7725634)... It's probably important to reflect again on the meaning and scope of book (Q571): do we mean book as a paper-based item or in a more general sense? What's then the difference between book (Q571) and literary work (Q7725634)? For example, I'm working on Pride and Prejudice (Q170583), both the properties were there, and I deleted instance of (P31)book (Q571). This is because that item it's a work in the broadest sense: when we have other editions and translations, they will all be linked to that. So, following the same reasoning, you should create a "book conservation status" property or even use both instance of (P31)lost literary work (Q1585442) + instance of (P31)book (Q571) (I'd accept it, because one statement informs about the WHAT of the item, and another about the STATUS of the item. M2c. Aubrey (talk) 12:15, 25 August 2016 (UTC)

Tool for importing Book references from ISBN

Hi,

I remember discussions, on Project Chat or other discussion places, about a tool/script/gadget, to automatically retrieve book info from online databases (such as LCCN, BNF, Worldcat), by just inputting an ISBN.

Is this already existing, or was it just some dream about what could be done in a (hopefully near) future ? --Hsarrazin (talk) 18:10, 22 August 2016 (UTC)

It was a dream. Right now, we are still struggling to understand how many items per book we need to create (especially regarding the relationship between Wikidata and Wikisource). ISBNs are related to editions of books, and when books are important enough to be put on Wikipedia they often have many editions or translations. So if we'll ever have that tool (which I think it's interesting) we should pay attention to this. Also, we need to understand which books we need to put on Wikidata. Right now, I think that we should at least have all the Wikipedia articles about books (which are already here) and the Wikisource ones (which are here but completely without statements, which is not nice :-)). Aubrey (talk) 12:18, 25 August 2016 (UTC)

Wikiproject Books 2.0

Assumptions and issues

Me and @Nonoranonqui: have worked all day on Pride and Prejudice (Q170583) (and relative pages) to better understand the relationship between a book which has numerous editions, translations, and even derived works as movies and plays.

We are going to use it as an example, as we still have some doubts that should be discussed by the whole community. As a premise, it's important to note that @Nonoranonqui: is a professional librarian, and I do get paid in my daily job to work with bibliographic metadata. Both of us participated in the last Wikicite conference, in which we discussed with dozens of wikidatans, librarians and information professionals about these things. We also had several in person meetings with Wikisource people both at the Wikisource conference last year and this June at Wikimania. This is just to explain that some of our assumptions have been discussed elsewhere, in different occasions with different people. We have some strong opinions on few things, while other things are much more open problems (at least, for us). As you can see in this very discussion page, there are recurring questions and doubts, and we would like to start to propose some solutions.

Assumption 0: books are complex, and at least we need 2 different levels in Wikidata to deal with them. As I said few months ago, we don't need to catalog the universe, but should strive for an effective catalog of books here that are primarily related to Wikimedia projects: we need to have items on Wikidata that are linked to Wikipedia articles, Wikisource transcriptions, Wikiquote quotes. Of course, we can put other books here on Wikidata, but IMHO we should always pay attention on the goal. Also, on the fact that we must maintain our "cataloging" the simplest as possible, because it's better for everyone :-) (for us and for the new users). The idea is that we must find a balance between a detailed way to put metadata on books, and the fact that we have to that in a distributed, non organized, dynamic manner, and maintenance is a hard.

Assumption 1: Wikidata is complex. Adapting Wikidata and books is and will never be a perfect match. We'll need to make choices, and every choice has consequences. After talking a lot with people much knowledgeable than me at Wikidata (as @Tobias1984:), I learned that some redundancy of properties is permitted (that is, putting some info in one item that is also in the superclass item), but it's not possible to put all the info in all the items. For that, there are queries. We must design our system thinking about queries, making sure that we have queries for finding books, and to maintain all the stuff related to books. I'll leave the explanation of the scaffolding principle to @Tobias1984:: to put it simply, it's better to have many items and few properties instead of few items and many properties.

Assumption 2: We must create an item only and just only when we need it. In pure wiki-style, create hundreds of thousand of items (both "work" and "editions") it's not necessary. If we have just a Wikipedia article, it will link it's own, single item. If on the other hand we have also Wikisource editions, it's better to discriminate and create other items. Of course, we must create a documented workflow, create instructions: what to do when I need to split a unique item in 2 "work" and "edition" items? Which properties must be used in the "work" and which in the "edition"? and so on. Unfortunately, in our opinion is difficult to define a priori kinds of items: meaning that it's probably not wise to explicitly say that an item is a "work", in the FRBR sense... Somehow, we think, it's the sitelinks to Wikisource, Wikipedia etc..., that define the item itself (if it is a "work" or an "edition"). If an item is linked to all the Wikipedia articles, well, we can assume that is on a "work" level. When we link it to Wikisource specific editions, we know we are talking about those specific editions. On the other hand, it's true that to agree on a property to be used for the "work item" or "edition item" (e.g. original language of film or TV show (P364) vs language of work or name (P407)) could avoid inconsistencies in the DB. It's another choice to make.

So, back to work.

One item can be a "work", in the sense of general, creative work. One example is, indeed, Pride and Prejudice (Q170583). From that root, we can create other items that are editions or translations of it. One example is Orgoglio e pregiudizio (Q26696290), an Italian translation that is on the Italian Wikisource.

That "work item" needs to have few properties (related to genre, characters, author), but generally no identifiers: we can put identifiers only if we know for sure that they are related to the work, and not to a particular edition.

One of these identifiers Worldcat Work ID (different from OCLC control number (P243), which does not exist yet.

• ex. http://worldcat.org/entity/work/id/3535, which can be found in many books in Worldcat, as a value in the schema:exampleOfWork property (click on "Linked data".

The "work item" can be linked to the Wikipedia articles (as it is now). But for Wikisource, it's probably better to connect each of the Wikisource books to single, dedicated items (which will be "editions"). On the Italian Wikisource, we also created a "Opera" namespace, a dedicated namespace for "works" (it's probably the best approach).

As you can see from the history of Pride and Prejudice (Q170583), we removed several statements that were about single editions of the work, and not the work per se. We also removed the property editions, has edition or translation (P747). IOHO, this property should be deleted or at least not used in "work items". There is an inverse property, edition or translation of (P629), much more useful for us. We can put edition or translation of (P629) in all the "edition items" and don't put anything on the "work item".

instance of (P31) is very tricky, both for "works" and "editions". We have it used with:

• book (Q571)
• literary work (Q7725634)
• work (Q386724)
• version, edition or translation (Q3331189)

I've often seen "work items" with instance of (P31)work (Q386724) and ""edition items" with instance of (P31)version, edition or translation (Q3331189), but that's not probably correct. Because people use the words "work" and "editions" in a different scope.

book (Q571)s are "paper based", but are used everywhere.

I don't know if it's reasonable (and a good idea) to use more than one value for instance of (P31): I don't know if it spoils the query (in a sense, it does, because it duplicates elements...). But we should discuss that.

IOHO, we should use literary work (Q7725634) for "work items", and book (Q571) for "edition items": it's simple and makes sense enough.

First issue: editions and translations have the same property. edition or translation of (P629). Not a problem per se, but at least we need to create some query to understand if we can retrieve all the single translations or edition in a certain language, if we can discriminate the former from the latter. In other words, if the same property is not generating noise.

Second issue: do we need to interpret this "work item" as the first edition? In that case, in the same item we could add publication date (P577) and even images or scans of the first edition. It's a reasonable choice, but it bears some heavy consequences (like, if I want to put some statements about that first edition, it maybe be better to have another dedicated item...). Also, related to instance of (P31).

We have 2 properties for languages: original language of film or TV show (P364) and language of work or name (P407). original language of film or TV show (P364) is perfect for "work items", while language of work or name (P407) for "edition items".

But we don't have the same luxury for dates: we only have publication date (P577). If we had another property, we could probably put publication date (P577) for "edition items" and the "date of first publication" in the "work item". Not sure if this is the solution (maybe it's to *delete* the double language property?), but we should discuss it.

There are probably many other issues, but it's enough for today :-)
Please read carefully few times before discuss ;-)

Nonoranonqui (talk • contribs • logs) and Aubrey (talk) 14:59, 25 August 2016 (UTC)

Update 08:44, 26 August 2016 (UTC): sorry for replying here, but it's probably better that at least my discourse and Nonoranonqui (talk • contribs • logs)'s stays in the same place. I wanted to reply to @Jane023: message below.

In Wikicite conference in Berlin, we discussed a lot with Wikidata staff and librarians about the 2 main ways to add a book:

• 1 new item
• 1 new statement with a lot of qualifiers

Both have pros and cons. But, mind you, at the moment, as we wrote before, we are not saying that we want to put in Wikidata all editions and translations of a book, but just those we need. And with need I mean: books we have on Wikipedia, Wikisource, Wikiquote, or even Commons. The idea is that we have a complex enough model to let Wikisource have a proper item for every edition it possess. Things gets complex enough pretty quickly: you can an Hamlet article in 100 wikipedia (and connected to 1 single item, the "work item"). But also you can have 2 or 3 versions of the book on English Wikisource (each should have an item). And you can have 4-5 translations in other Wikisources (each, too, should have an item). And I'm not even talking about movies, plays, tv series; each of those *already* have a Wikipedia article in *at least* the English Wikipedia.

This is why I think that we should go with the "create an item when you need it" rule. Aubrey (talk) 08:44, 26 August 2016 (UTC)

Let me elaborate a bit more on that.

Create an item when you need it rule

This is crucial and is in two parts, namely adding information to an existing item, or creating new items: to attract new contributors, you need to first specify the bare minimum to add to an item that is a book. So if a Wikidatan wants to flesh out an old book item imported in 2013 that currently has zero statements, what is the minimum you need to pick it up for this project? Hint: Maarten Dammers uses a basic set of statements for paintings that he imports from museum collections. These are "instance of", creator, collection, inventory number, collection title (used as label in the collection's language). Do you have a basic list like that for books? Jane023 (talk) 13:07, 26 August 2016 (UTC)

Discussion

• I've read carefully all the text and I agree with every single word. About the issues: in my opinion we should divide properties for editions and translations; I think "work items" can indicate the first edition too; about date of publication I think a single property is fine, we can assume that in items with instance of (P31)literary work (Q7725634) it refers to the first edition, while in items with instance of (P31)book (Q571) it refers to that edition. --Epìdosis 18:48, 25 August 2016 (UTC)
• My personal opinion is as follows,

1. has edition or translation (P747) should be used only in literary work (Q7725634) or first edition.
2. There should be a separate property called translations which also should be used only in literary work (Q7725634) or first edition.
3. edition or translation of (P629) should be divided into edition of and translation of, which should be used in the editions and translations only.
4. publication date (P577) can be renamed as date of first publication
5. There should be a new property called publication date of edition -- Bodhisattwa (talk) 19:36, 25 August 2016 (UTC)

No, this is just creating a bunch of new properties and this will lead to more difficulty to query data because we increase the risk of not using the good property in the right item. Snipre (talk) 20:44, 25 August 2016 (UTC)

Hi @Snipre:, whom did you reply to? I don't understand :-/ Aubrey (talk) 08:27, 26 August 2016 (UTC)

Aubrey My comment was for the proposition of Bodhisattwa. Snipre (talk) 11:36, 26 August 2016 (UTC)

I think you can add multiple dates of publication and add qualifiers for the editions, such as ISBN numbers or whatever. I don't see why editions need separate item numbers unless this is significant somehow in the content of the work. For older editions, there can be huge descrepancies in language editions, and maybe for important rare works you need a separate item per collection. Jane023 (talk) 22:29, 25 August 2016 (UTC)

@Jane023: well, technically you can add multiple date and use qualifier but that's not a good idea. There is a lot of reason why we need to separate works and editions. The first and more important, is that it's more readable for both human and bot/tools (qualifiers are great but when all the values are duplicating, it could be tedious to read, and I'm only talking of 2 editions, some books have hundred thousand editions) ; readiblity is very important because it's linked to many others things like contribution, correction and improvement (users are more likely to edit small easy-to-understand chunks than big-fat items). An other example Wikisource-specific, if you only have on item, how do you do the interlink beetween all the editions (spoiler: you can't ; and BTW, the enws interwikilink seem wrong to me). And last but not least, that's all the GLAM of the world do for centuries (and it's not an just an argument from authority, it's more pragmatic: a system so widely used may not be perfect but it very probably not bad either). Cdlt, VIGNERON (talk) 09:02, 26 August 2016 (UTC)

• This is a great summary of the issues; thank you for working on this! There's lots that needs figuring out, but I think there's a few solid bits already that we mostly seem to be starting to agree on: the work-item shouldn't have any edition-identifiers; queries can get us a lot of the output we want, while leaving the datamodel simple and understandable; Wikisources should connect to edition-items (only disambiguation pages should like to work-items). Some things I've wondered about: why is version, edition or translation (Q3331189) a subclass of book (Q571) (e.g. are there also editions of films, music)? What is the top-level subclass of (P279) of everything on Wikisources (this would be cool, for finding e.g. things that are missing from Wikisource but are eligible to be there)? Maybe it is literary work (Q7725634), but there's more than than just those on Wikisources.

  Anyway, it is very good advice to remember the end goal when entering books' info in Wikidata. For me, one end goal is "a means to browse Wikisource material (in all languages) and search by all the metadata possible" — this will mean we can build tools to get the Wikisource catalogue incorporated into third-party meta-catalogues (e.g. the National Library of Australia has a cross-institutional catalogue that would probably ingest an Australian-content feed if we had such a thing).

  I'm ignoring all that for now because I want to finish scraping all Wikisources for their data! :-) Which is jolly fun and fraught with annoyances (like DNS resolution failing intermittently on WMF labs?! ergh.) Samwilson (talk) 03:30, 26 August 2016 (UTC)

1. @Aubrey, Nonoranonqui: The great essay!
   • Since LUA has not able to search items by property value, we still need has edition or translation (P747), not only edition or translation of (P629).

Could you elaborate, please? My concern is that it's impossible to maintain has edition or translation (P747), while edition or translation of (P629) it's much much easier. Put it simply, when you create a new edition item (it could be a book, but also a movie, a play, etc.) you use edition or translation of (P629) in each. In each, you use also instance of (P31) with movie, book, play, etc. I don't see the fault in this idea, but I'm maybe missing something.

Let Qxxx has edition or translation of (P629) with value Qyyy. How to a LUA-code called from Qyyy-linked page can find Qxxx? Todays the only way is to have reverse property on Qyyy. E.g. how to find French translation (Q17352355) of Misery (Q3539674) (s:ru:Тоска (Чехов)) without using has edition or translation (P747)? -- Sergey kudryavtsev (talk) 10:06, 29 August 2016 (UTC)

Reasonator is able of finding the editions... (see "From related items" in Q3539674). --Edhral (talk) 11:47, 4 September 2016 (UTC)

1. • Some items use publication (Q732577) as P31's value.

:-(

1. • There is third issue — original language of film or TV show (P364) vs language of work or name (P407) in a original (not translated) work. You use original language of film or TV show (P364) = English (Q1860) in Pride and Prejudice (Q170583), not language of work or name (P407). Why?

I understand now what you mean. The point is: when do we have to use original language of film or TV show (P364)? Either we use that in all the "edition level" items (everytime, we say that a translation has a language and an original language) or we use it one time, in the first, root, "work item". I think that, query-wise, you can always find which was the original language. It's maybe even redundant to have a double property. I really don't know how to solve it, but putting original language of film or TV show (P364) in the root "work item" seems to me a good idea. Aubrey (talk) 09:02, 26 August 2016 (UTC)

You don't understood me. If i want to determinate writted language, in which writted some work, then first i should look for language of work or name (P407) (which have translated works). If P407 is absend, then i should look for original language of film or TV show (P364) (for not translated works). I prefer to have language of work or name (P407) as writted language for any work! This is a polymorphism principle of object-oriented programming (Q79872). -- Sergey kudryavtsev (talk) 10:33, 29 August 2016 (UTC)

-- Sergey kudryavtsev (talk) 05:12, 26 August 2016 (UTC)

• I totally agree, and I like to stress on the two assumptions. The world is complex and there always be exceptions and strange cases (for instance, the Bibliothèque nationale de France ID (P268) is for edition, and for works, and for people, and for places, and for concept, and for words, etc.) that doesn't perfectly fit the structures and it doesn't question the structure itself which seems the best we can possibly do. Cdlt, VIGNERON (talk) 08:41, 26 August 2016 (UTC)
• Don't forget the manuscripts (one big reason not to mix up works and first editions)! We have some thousand Wikipedia articles and Commons files describing/depicting manuscripts, only a few of them are autographs. --HHill (talk) 09:14, 26 August 2016 (UTC)

Jakob's thoughts

Thanks alot for this well thought summary, especially the basic assumptions. I have feedback on three issues:

• Yes, works should only have work identifiers, such as LibraryThing work ID (P1085). Unfortunately some identifier properties can be used for both, work identifier and edition identifier (e.g. GND ID (P227)), this has to be documented very well.
• Many works only have a single edition. I'd object having to create two items in this cases. One item for both work and edition (and possibly physical as well, for instance manuscripts) is enough as long as there are no other more editions or translations.

Yes. Many books just have a Wikipedia article connected to one WD item, some have just a Wikisource transcription (connected to another Wikidata item). It's ok to have just one item when you don't need two. This is why I added "Assumption 2". Aubrey (talk) 12:53, 26 August 2016 (UTC)

• Don't overestimate instance of (P31). Edition items are exactly those items that make use of edition or translation of (P629), no matter what they are instance of. I suppose that nothing should be edition of another edition but reality is always more complex than expected. The following query seems to reveal many errors:

SELECT ?work ?workLabel ?wtf ?wtfLabel ?edition ?editionLabel WHERE {
    ?edition wdt:P629 ?wtf . ?wtf wdt:P629 ?work .
    SERVICE wikibase:label { bd:serviceParam wikibase:language "en" . }
}

Try it!

-- JakobVoss (talk) 12:11, 26 August 2016 (UTC)

True. This is why I like edition or translation of (P629): in a why, it just identifies "edition items", and that is nice and simple. The issue for me is about how to handle languages (original and translation) and dates (original and translations or editions). Aubrey (talk) 12:53, 26 August 2016 (UTC)

@JakobVoss: Vielleicht schon erledigt, aber nur der kurzen Hinweis, dass sich die Eigenschaft GND ID (P227) ausschließlich auf Normdaten (GND, Typ w = Werke) bezieht. Für einzelne Ausgaben (= Editionen) gibt es, wie auf der Projektseite beschrieben, DNB edition ID (P1292). Gruß --Kolja21 (talk) 20:09, 19 June 2017 (UTC)

Finn's thoughts

Here are a few thoughts:

1. book (Q571) is bad and should not be used for bibliographic information. This is because a book is more general than other printed/published literary works. For instance, a book might be be a notebook with blank pages. I have used "book" but I think it would be better with literary work (Q7725634) and (which I did not know before) version, edition or translation (Q3331189).
2. Most Wikipedia articles about literary works often are more like about the superwork (Q24261960) (the original work, the translation, music, recordings), see, e.g., Stille Nacht Heilige Nacht (Q172152). How do we deal with that?
3. We should have an item for individual books rather than book (Q571). For Book of Kells (Q204221) it is codex (Q213924) and illuminated manuscript (Q48498).
4. I would tend to agree with User:JakobVoss about "Many works only have a single edition. I'd object having to create two items in this cases. One item for both work and edition (and possibly physical as well, for instance manuscripts) is enough as long as there are no other more editions or translations." But what if later editions appear?
5. An edition might contain a new work. For instance a new edition of a classical novel might contain a new foreword by another author.

— Finn Årup Nielsen (fnielsen) (talk) 14:33, 26 August 2016 (UTC)

For 2. I don't see this as a problem but hairsplitting. superwork (Q24261960) is no generally agreed-upon concept but it was just created by you.

For 4. we could

either make these items instance of both literary work (Q7725634) and version, edition or translation (Q3331189) and check that the following is not allowed:

SELECT ?wtf ?wtfLabel ?edition ?editionLabel WHERE {
    ?edition wdt:P31 wd:Q3331189 . ?wtf wdt:P629 ?edition .
    SERVICE wikibase:label { bd:serviceParam wikibase:language "en" . }
}

Try it!

or make these items edition or translation of (P629) of themselves and check with the query given above in my first comment. -- JakobVoss (talk) 15:25, 26 August 2016 (UTC)

book (Q571) ("book") is not perfect, but literary work (Q7725634) is no solution since it excludes many types of books. Photo books or catalogs for example are no literary works and many books aren't an "obra creada por un escritor". --Kolja21 (talk) 20:29, 26 August 2016 (UTC)

I created a more general item, a superclass of "literary work". Not a fundamental problem then. I think we talked about this on Will you be its creator? click here. a while ago - don't forget here to work in collaboration with them, they definitely use literary works as well, and all of them aren't printed editions. author  TomT0m / talk page 15:39, 30 August 2016 (UTC)

It's written or drawn work (Q25839930)   . author  TomT0m / talk page 15:57, 30 August 2016 (UTC)

P31, FRBR and VIGNERON maybe crazy idea

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

I see that there is a lot of discussion around instance of (P31), and right now we use it to store the FRBR level but when I re-look at it, I wonder: is it a good idea and shouldn't we have a FRBR level property? I won't probably solve the book (Q571) issue (who date back to the beginning, see Talk:Q571) but I think it could help. It's adding yet-another-property (and I'm not a big fan of multiplying properties) and would be maybe a bit redundant but I believe it can help to check more easily other properties consistency (and to focus more quietly on how to deal with instance of (P31) values). Has it been discussed before? (I don't recall it) Cdlt, VIGNERON (talk) 09:48, 27 August 2016 (UTC)

+1. Indeed it looks like instance of (P31) will not solve the problem. I still hope one day Wikidata can replace w:Template:BibISBN. If we want to use WD as database for books and articles that are cited in Wikipedia we also need a property like "cited in WP" connected with a type of protection mechanism. Basic bibliographic datas need to be save. --Kolja21 (talk) 10:33, 27 August 2016 (UTC)

Shonagon's feedback

• Thanks Aubrey and Nonoranonqui for this subject.

I participated in editing travel books : Shonagon/WikiProjet_Récits_de_voyage

and French editions of Homer's Iliad and Odyssey

• > Assumption 0: books are complex, and at least we need 2 different levels in Wikidata to deal with them.

> Assumption 1: Wikidata is complex. Adapting Wikidata and books is and will never be a perfect match.

> Assumption 2: We must create an item only and just only when we need it.

Yes

• > "work item" needs to have few properties, but generally no identifiers

Some organisations like the Bibliothèque nationale de France (Q193563) are creating IDs for works. It's a big work in progress but it has begun and can be already used. For example, movies based on Jules Verne's work with BnF identifiers (there is not A Trip to the Moon (Q244971) because there is not yet Id for the work From the Earth to the Moon (Q53592) in BnF catalog)

• > We also removed the property editions, has edition or translation (P747). IOHO, this property should be deleted or at least not used in "work items".

In theory maybe it should be. In practice édition has edition or translation (P747) is a so much useful property that it's difficult to image editing without it. There is indeed a transitive relation (like with has part(s) (P527) and part of (P361)) which makes this property logically useless but thinking that contributors would made searches and/or requests to retrieve the edition of literary work (Q7725634) is really the worst way, long and complex. Having the editions in literary work (Q7725634), has the other advantage to show to newcomers on an item's page, what is done, if there is the need and where it could be improved. Maybe not having has edition or translation (P747) could be resolved with interface or gadgets, but for now we need it for contribution.

• > IOHO, we should use œuvre littéraire (Q7725634) for "work items", and livre (Q571) for "edition items": it's simple and makes sense enough.

Pro. I didn't edit like that but IMHO it's a really better way, as you explained.

• > Second issue: do we need to interpret this "work item" as the first edition? In that case, in the same item we could add date de publication (P577) and even images or scans of the first edition.

It is not a general case. It's often the way for many literary work (Q7725634) but it fails with old literary work (Q7725634) which don't have first publication date (P577) and many books have been published a long time after the work was created. Date of first publication will not suit to those cases. inception (P571) is a property which suits for literary work (Q7725634), as recommanded in RDA (cf [(6.4) Date of work]), application of FRBR's logic, and as it was adopted with reason in WikiProject Books.

• > original language of film or TV show (P364) is perfect for "work items", while langue de l'œuvre (ou du nom) language of work or name (P407) for "edition items".

Yes. That's the way I use. --Shonagon (talk) 20:43, 23 September 2016 (UTC)

Silvio's comments to Books 2.0

OK, that's a very interesting discussion about very specific and concrete scenario - that imposes clear constraints. Which is good. I apologise for not being part of WikiCite 2016 nor participating to any discussion on this topic so far. I hope my comments may help the discussion though - I'm technically bias by my knowledge of FRBR, which I repute a great standard and used in my work (see SPAR Ontologies, http://www.sparontologies.net).

Works and editions

First of all, I found the bipartite description of books by means of the intuitive concepts "work" and "edition" a quite good (and necessary) compromise. While I like simplicity, I'm also a lover of a clean design of a domain that should be less ambiguous as possible.

While I think the use of literary work (Q7725634) for identifying works is fine, I'm quite worried to use book (Q571) for editions. In Wikidata, the concept book is used for describing a quite precise medium, but it is restrictive for particular editions. I know that the focus should be only on books so far, but even WikiSource contains other kinds of entities such as papers - that are not book and, maybe, can be part of a book. I think there can be several other exception of this kind in the future, and thus I would suggest to avoid to use book (Q571) for editions, while preferring a more generic edition (Q3331189).

Mixing up disjoint concepts

My additional worry is about mixing up the concept literary work (Q7725634) with edition (Q3331189) like suggested for handling the "first edition" issue. Formally speaking, any concept (or class) is usually characterised by a set of property that formally distinguish it from another. Literary works and editions are two related but distinct kinds of objects, and saying that for one specific instance of a "book" they coincide (i.e. for the first edition) is is just odd.

However it would be important to clearly identify what was the first edition of a literary work. Note that we cannot use the publication date for inferring this, because it is not always true that the edition with the oldest publication date is the first edition - e.g. I could not have all the data available in Wikidata about a particular literary work. Thus, a "first edition of" property (sub-property of "edition of") and publication dates can be used for answering to two different kinds of query:

1. What is the first edition of a certain literary work?
2. What is the oldest edition Wikidata has available of a certain literary work?

Note that 2 does not imply 1, while 1 implies 2.

In addition, having a "first edition of" property would simplify also the identification of possible inconsistency (e.g. the presence of two "first editions" in Wikidata) without touching the main literary work they refer to.

About editions and translations

I totally agree that two properties, "edition of" and "translation of", would be necessary for clearly distinguishing between editions and translations.

However, I would have some terminological concerns here. Intuitively (please librarians correct me!) a translation is actually a particular edition that has been derived (i.e. translated) from another edition. As far as I know, a translator does translate a particular published (or will-published) version of a literary work, does not she?

[Clarification: in my mind, a literary work is used to identify the abstract notion of a certain book, while its actual content (the text written by the author) is a property of the edition]

Thus, my point is that while the property "edition of" goes from editions to literary works, "translation of" should go from editions (the translated version) to editions (the original language version). In addition, a translation is also an edition of a certain literary work, according to my premise above.

BTW, with this approach one can also track the (quite odd, honestly) scenario where a translation of an edition has been derived from another translation :-)

Publication date and languages

Since there is the need of maintaining controlled the number of properties used for describing the (meta)data of a book, I really don't get the point of having pairs of properties related to publications and languages for both literary works and editions. From my personal perspective, these are attributes of editions - that already allow one to perform any possible query, if all the aforementioned suggestions are considered.

For instance, we can ask for:

• the first publication date of a literary work by looking at the publication date of its first edition (by means of the property "first edition of");
• the language used for the first edition of a literary work by looking at the language of its first edition (by means of the property "first edition of", again);
• the languages in which a literary work has been translated by looking at all the edition of such literary work that are translation of another edition;
• and the list may go on.

Just two properties are needed then, i.e. the already existing publication date (P577) (substituting that "is first published" with "is published" from its definition) and a new "written in" property, both associated to editions.

[As a side note: the property language of work or name (P407) is rather ambiguous, since it can be used both for works and people. In this cases I would suggest to create always two properties or, if this is not possible, to generalise the property name in a way that seems more generic – even if this generalisation is not possible in this specific case, I think]

A diagram

I've prepared a diagram at http://www.sparontologies.net/static/literary-works-and-editions.png (I couldn't upload it in the wiki, I don't know why...). The diagram devised in Graffoo (http://www.essepuntato.it/graffoo) describes graphically all the modification I've proposed in the aforementioned points. The property indicated with "[new]" do not currently exist in WikiData.

An example according to the diagram

Pride and Prejudice (Q170583) 
	instance of (P31) literary work (Q7725634)

Pride and Prejudice (1813)
	instance of (P31) edition (Q3331189)
	first edition of [new] Pride and Prejudice (Q170583)
	publication date (P577) "1813-01-28"^Gregorian
	written in [new] English (Q1860) 

Orgoglio e pregiudizio (1945) (Q26696290) 
	instance of (P31) edition (Q3331189)
	edition of [new] Pride and Prejudice (Q170583)
	translation of [new] Pride and Prejudice (1813)
	publication date (P577) "1945"
	written in [new] Italian (Q652) 

Characters names

This is just provocative, but some editions (e.g. Harry Potter in Italian) have changed the actual names of almost all the characters of the books. Thus, in principle, a bipartite description could be needed also for characters.

Essepuntato (talk) 12:00, 29 August 2016 (UTC)

@Essepuntato: I don't understand your system: so if I have a translated version of a work, I have to create one item for the translated version, one for the first edition in the original language and one for the work even if I don't need the item for the first edition in the original language ? Why do you need to increase the number of items when two or even one if necessary ? Snipre (talk) 14:41, 29 August 2016 (UTC)

@Snipre: No, you don't. The example I've done before was just a complete one. However, one is free to create only the items that are really needed, e.g. you can create the item for the translation of a work that is directly linked (via "edition of") to the actual literary work. But this would need a clear characterisation of what we mean with "literary work", "edition" and "translation".

In the model I've proposed above, I use "literary work" for speaking about the work per se, without referring to a specific version of it - e.g. when I talk about The Betrothed by Alessandro Manzoni with a friend. Then, with "edition" I refer to a particular "crystallisation point" of the text of such literary work that has been published - e.g. for The Betrothed we have two important editions: 1827 and 1842. Both of them, of course, are in Italian. Then we have translations, i.e. particular editions where the original text of a particular edition has been translated into another language (e.g. the English edition of The Betrothed published in 1828 and entirely derived from the text in the 1827 Italian edition). From this perspective, then, translation are particular edition of a literary work, that happens to be derived from another edition in a different language.

Having that model, it is entirely possible, thus, to have only partial data, e.g. the item for the literary work and just an edition of it (that may not be the first edition and could be written in a language that not necessarily is the same of the first edition), e.g.:

The Betrothed 
  instance of literary work

The Betrohed (1828 - English)
  instance of edition
  edition of The_Betrothed
  written in English
  publication date "1828"

However, I can see the need of have some sort of duplication of the data between literary works and editions - in particular if I want to create only a literary work without caring about specific editions. In this way, the approach with "first publication date" and "original language" could be a good option indeed. However, I would discourage to say that an item is, at the same time, the literary work and the first edition. If a "first edition" item is really needed, then we should create it explicitly and link it to the related literary work in some way (see my proposal of using the "first edition of [new]" property above).

@Essepuntato: Are you kinding me ? How can you propose a complex system to just say after that contributors should use a partial version of it ? Partial models are just the worst thing we can use because this means that nobody will never use the complete model, it's called the law of the least effort. So if your partial model can do the work of the complete model, say it and forget the complete model. Snipre (talk) 22:27, 29 August 2016 (UTC)

@Snipre: The fact that the system is complex (even if, I have to say, it does not look more complex to me than what you have proposed) does not mean that we need always to specify all the data that the system can describe. If some data is not known as in the example above (where, for instance, I suppose not to know what is the first Italian edition from which the English one has been translated), then I can always create the items I need without adding necessarily also the first edition item. In this sense one can use "partially" the system: when data are not available for creating a particular item, I use only a reduce set of properties and types. Essepuntato (talk) 06:27, 30 August 2016 (UTC)

Some thoughts by Snipre

1. Splitting edition and translation is just a wrong good idea: you can have several editions of one translation. So translations should be considered in the same class as editions in the original languages.
2. original language of film or TV show (P364) and language of work or name (P407) have to be replaced by P2439 (P2439): the distinction between original language and language is not relevant. The original language is the language of the work. So to find the original language of an edition, just use the language from the work item. Then you have a similar pattern for all editions and works with an unique property. Same for publication date.
3. With the structure described below we can have one item for both work and edition in the case of a work with only one edition. But we have to add instance of (P31):work (Q386724) and instance of (P31):version, edition or translation (Q3331189) in this item, and edition or translation of (P629) with the Q number of the item as self reference.
4. The distinction between work or book is depending of the will to allow movies or theatre plays or other kind of representation to be connected to the work item used for books/editions.

Work	Edition1	Translation
instance of (P31):work (Q386724)	instance of (P31):version, edition or translation (Q3331189)	instance of (P31):version, edition or translation (Q3331189)
-	edition or translation of (P629): work item	edition or translation of (P629): work item
P2439 (P2439)	P2439 (P2439)	P2439 (P2439)
publication date (P577)	publication date (P577)	publication date (P577)
-	edition number (P393)	edition number (P393)
author (P50)	?	?

¹ edition is defined here as an edition in the original language only

Work and edition in one item
instance of (P31):work (Q386724) instance of (P31):version, edition or translation (Q3331189)
edition or translation of (P629): itself item
P2439 (P2439)
publication date (P577)
edition number (P393) = 1
author (P50)

• the original language is the language of the work item connected to the edition item using edition or translation of (P629). No need to indicate it in the edition items.
• the original publication date is the publication date of the work item connected to the edition item using edition or translation of (P629). this is the same publication date as the one indicated in the first edition of the work in the original language, but as we can't ensure to have an item about this edition we have to add that information in the work item.
• the identification of the translations is done by comparing the language of the edition with the language of the work item. All edition items with a different language than the one in the work item is a translation.
• we can go further and discuss about the need to add author (P50) to edition/translation items.

We don't need to create new properties: we can extract a lot of information using inference. Snipre (talk) 13:07, 29 August 2016 (UTC)

@Aubrey: You should not focus only on books or literary work but include movies and theatre plays too.

How do we have to consider the relation between these items:

* Pride and Prejudice (Q170583)

* QXXX, book, first edition in English

* Orgoglio e pregiudizio (Q26696290), book, Italian translation

* Pride and Prejudice (Q3356244), English TV serie

* Pride & Prejudice (Q270385), movie

In some point we should be able to connect all these items through an unique item, the superwork item decribed above. But finally shouldn't we have only one work item for movie, books, TV serie,... and avoid the intermediate classification level literary work ? Snipre (talk) 15:22, 29 August 2016 (UTC)

Some thoughts by Aubrey

I've been conflicted by Snipre, so I'll add my thoughts here. (it's really difficult to discuss about these things :-() Thank you @Essepuntato: for you great effort, it's really precious. I'll try to reply following your structure. Works and editions: I kinda agree with you. I like literary work (Q7725634) for works that started as books (like Pride and prejudice), and we can try to handle eventual exceptions (like suggested by @Kolja21:) like photo books or similar adding other properties. A doubt I still have is: it's a good idea to use multiple values with instance of (P31)? Does it pollute queries?

@Aubrey:, while having multiple values is always possible, I would discourage to use both "literary work" and "edition" in this context. A literary work item is one thing, an edition (even the first edition) of that item in another. Thus, I'd suggest just to specify one type per item here. Essepuntato (talk) 06:29, 30 August 2016 (UTC)

Because if we can, we can imagine to use instance of (P31) with books, edition, article, movie or whatever item it's best for the single case. For "edition items", version, edition or translation (Q3331189) could be always used, in addition with other items. Is it too complicated?

@Aubrey:, that would be a good option to me, and it is not so complicate from my point of view. Essepuntato (talk) 06:29, 30 August 2016 (UTC)

Mixing up disjoint concepts: having the "work" and the "first instance of the work" together is for simplicity and accessibility for the user. I'm worried that if we explode every book (or movie, etc) we see in 2 or multiple items, that will be very confusing for everyone. Of course, we can create a "first edition item" for those book when that feels necessary, but I fear that many people would create thousands of "first edition items" detached both from Wikipedia articles or Wikisource books... That is not bad per se, but we risk to have multple items even when that's not necessary. This is related to the fact that, for me, the "work item" is not necessarily a strict FRBR-work. If it was that, we should have very few properties. FRBR is extremely abstract and I'm worried people wouldn't just understand that. Also, crucially, FRBR is in the eye of the beholder. Pride and prejudice (the TV series) is a manifestation of the original work, but it's also a work per se. This is why I don't want the FRBR-level hardcoded in the item. I'm willing to accept some vagueness if it helps maintaining a simple, flat structure that is effective. We have many thousands of Wikipedia articles who are linked to a Wikidata item. Many are about the work in a more general sense, others also have a publication date and other more "edition-level" metadata. For me, it's kinda OK like it is, because having just one item for book is very simple. IMHO, we must used a more complex structure (like you proposed) if we have the need of other items (like Wikisource edition to connect, or, in the future, specific editions in Wikipedia bibliographies and references).

@Aubrey:, then a duplication of some of the most important metadata at a literary work level (first publication date, original language, etc.) is the only option to me, since, as I already said, specifying an item as literary work and edition is quite odd to me - see also the answer I've provided to Snipre above. Essepuntato (talk) 06:29, 30 August 2016 (UTC)

Editions and translations: as said before, I kinda like edition or translation of (P629), but I understand the advantages of dividing it in 2 different properties. Again, we should find good queries to retrieve both the items, if needed (I'm not a SPARQL master and get confused easily in what can be done or not done with queries). Aubrey (talk) 15:29, 29 August 2016 (UTC)

@Aubrey:, well, of course having just the property edition or translation of (P629) and the ones that allow the associations edition->language and literary work-original language is enough for inferring (via SPARQL) if one edition is a translation (which, in my mind, is a particular kind of edition) or if it is just another edition in the same original language of the literary work. However, having the two separated property would make everything more clearer to me - still, see the example above on The Betrothed I've proposed to Snipre. Essepuntato (talk) 06:29, 30 August 2016 (UTC)

Unfortunately (?) I'm going away for 2 weeks without a proper connection. Feel free to continue discussing without me :-D (I'd love to see some consensus when I come back :-D). Aubrey (talk) 19:51, 30 August 2016 (UTC)

<Head thunk>

OMG, no hope of me wading through all of that. This is going to need to be broken down to bite-sized chucks if you wish to have a broader conversation. A bicker over book <-> literary work <-> ... is conceptually challenging, especially when you add in fiction v. non-fiction; reports of boards; multi-volumed biographical dictionaries. Get your base right, then work from there. I think that you have to clearly separate out editions as they are conceptually different. [As a note, the literary work is what? I find it a nefarious and even more confusing argument, is it what the author gave the editor, or the publisher, the published work at its first edition, the copy in a person's archives ... <aagh>)

From a Wikisource point of view we need to conceptually get EDITIONS right. Each of our editions can come from so many sources, predominantly books (historically), though ever increasing base from the mid 19th century to the myriad now available. We need to make it clear and easy to have an addition added for a edition with the right and pertinent detail. The prompts for editions are still limited, and getting that fixed and extras included like price/number of pages/edition number/commons scan/wikisource index/...

So I suppose is there a clear separation between editions and books. I sometimes feel that the BOOK >> EDITION dominates the discussion so BOOK is seen as the leader, whereas EDITION has many lead-in components, and that needs to fit within Books/Newspapers/Journals/Documents/... There are many 19th century articles, or serials that became a book, etc. They have to be treated separately.

I will now shuffle away as it all does my head in. :-/  — billinghurst sDrewth 02:50, 1 September 2016 (UTC)

So what should one do?

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

I've read this discussion, which I find fascinating, and I am still no further along than I was when I started. Today, as a contributor, if I want to enter a book so that I can use it as a <stated in> reference, what is the best practice for modeling that book? I've been waffling between calling it a book (single item) and doing a work/edition pair. Whatever I do I seem to get constraint violations. Help? PKM (talk) 19:49, 19 June 2017 (UTC)

Can you be more specific? I make items for books all the time (mostly catalogs or reference works) and don't have any problems with constraints (but maybe I just don't notice them). Jane023 (talk) 21:13, 19 June 2017 (UTC)

I turned on the constraint gadget.  :-/

Survey of Historic Costume (Q29002765) fifth version, edition or translation (Q3331189) of Survey of Historic Costume (Q28946447). This is a book where editions come from different publishers in the US and UK, and later updated editions have a different co-author. Constraints on the edition:

• Entities using the subtitle property should be instances of work (or of a subclass of it), but Survey of Historic Costume currently isn't.
• Entities using the author property should be instances of work (or of a subclass of it), but Survey of Historic Costume currently isn't. Entities using the publisher property should be instances of work (or of a subclass of it), but Survey of Historic Costume currently isn't.
• Entities using the publication date property should be instances of one of the following classes (or of one of their subclasses), but Survey of Historic Costume currently isn't: work, sequence annotation, apostolic constitution, file, format, decree, folklore.
• Entities using the original language of work property should be instances of work (or of a subclass of it), but Survey of Historic Costume currently isn't.

And on the work:

• edition number is not a valid qualifier for edition(s) – the only valid qualifiers are: language of work or name, translator, volume, applies to part,publication date (so I guess this should be a property and not a qualifier of "edition"?) - PKM (talk) 22:06, 19 June 2017 (UTC)

Hi @PKM:,

There is known bug with the constraint gadget, it is unable to check more than 4 classes :( So the 4 first constraint violation are false-positive (edition is - obviously - a subclass of work), you can just ignore it.

For the edition number (P393) on the work, here it is e real constraint violation. I guess that the edition number (P393) is redundant and unnecessary but I'm not 100% sure.

Cdlt, VIGNERON (talk) 07:36, 20 June 2017 (UTC)

@VIGNERON: Oh, thank you! I'll ignore constraint violations on these for now. That helps. - PKM (talk) 19:29, 20 June 2017 (UTC)

Items without statements related to books

In several languages, frequently there are project related categories at Wikidata:Database reports/items without claims categories.

For English Wikipedia, e.g., there is now on Wikidata:Database reports/items without claims categories/enwiki:

• w:Category:Series of children's books 235
• w:Category:Series of books 167
• w:Category:Science fiction book series 113
• etc.

Adding at least one statement ensures these can be completed more easily. --- Jura 18:53, 30 August 2016 (UTC)

Some thoughts by Chiara

I guess I have to make some clarifications from a librarian’s point of view :-) even if probably will not influence the conversation:

• there are many examples of translations not from original work/original language but from another translation, as in the contemporary world (e.g. arabic books are usually translated from their english or french versions) as in the ancient world (many greek works were translated into latin, in Middle Age, through arabic version - and in the meantime the greek works have been lost);
• Like @Aubrey: said, Pride and Prejudice (film 2005 by Joe Wright) isn’t a manifestation of Pride and Prejudice (book by Jane Austen) but a new work because it has a different authorship responsibility.

In effect: https://en.wikipedia.org/wiki/Pride_and_Prejudice_%28disambiguation%29

So, why we should have a “superwork item”? It is a real question!

The item Pride and Prejudice (film 2005 by Joe Wright) will have the property “based on” Pride and Prejudice (book by Jane Austen);

• @Essepuntato: I totally agree with you! In particular on the importance of diversifying the work and the first edition

Nonoranonqui (talk) 09:08, 31 August 2016 (UTC)

@Nonoranonqui: The problem I see in your reasoning is the lack of symmetry: just take the example of a book written based on the scenario of a movie or of a video game: by applying your reasoning, you can't define an item with literary work for the work item and you have to follow the same structure that for movie and used “based on” property to link the book to the movie. I agree with you about the fact that we don't need a superwork item ... because we have the work item. But to have a coherent system we shouldn't define that item as literary work but only as work which can be used by all forms of manifestations independently from the original format.

Again we can't just create a system for books but we have to have an global overview of the different formats which can be used to express a work, then define a relation structure between the different formats and finally we can go back to the book format and define how we want to structure that subsystem.

Just for example we have Halo: The Fall of Reach (Q2453886) which a book from a book serie Halo (Q3126185) based on the video game serie Halo (Q1747150). Can we really create a work item for that book defined as literary work ? What is the relation between the book and the video game: "based on" property ? Snipre (talk) 09:53, 31 August 2016 (UTC)

There is also items for fictional universes. This seems relevant to model the "Halo Universe" and the works that describes and take part in its. But I agree there is a real question : if the story of a book is the sequel of a story of a video game, for example. Narrative follows narative, whatever the support/manifestations are. Class of manifestation actually for a videogame for exampe, because a video games can be played a lot of times, each is an instance or a manifestation of the work. That's a totally valid argument to separate an edition from the work, for example. author  TomT0m / talk page 12:37, 31 August 2016 (UTC)

Sorry,@Snipre:, maybe I haven’t explained myself. IMHO, we need a “work item” for each kind of work: one for “work-literary work”, one for “work-movie work”, one for “work - play (theatre) work”, another for “work-video game work”..and so on. Each “work item” could have its editions or translations or performances...and, then, we can link work-items with each other e.g. with the "based on" property Nonoranonqui (talk) 14:07, 31 August 2016 (UTC)

@Nonoranonqui: Ok, this seems more structured but perhaps a little more confusing for other format than books. If the structure work-edition/translation works fine for books I am not sure that find a similar interest to use it for movie. To discuss. Snipre (talk) 14:14, 31 August 2016 (UTC)

It seems to me that you are creating some false structure. There will be a primary/seminal work, and that is what it is, be it a book, a movie, a scrawling on a wall. From that primary work, there may be editions, translations, based-ons, etc. A certain translation may be based on the fourth edition, and that simply should reflect that aspect. A certain later Afrikaans film could be based on a Dutch translation of a German work, and it should follow the trail, not be against some artificial structure or concept, or the creation of some new individual's conceptual view.  — billinghurst sDrewth 02:19, 1 September 2016 (UTC)

@billinghurst: I don't think Nonoranonqui is introducing a false structure here. When you talk about the need of having a "a primary/seminal work", what we you are actually talk about is a "a primary/seminal edition of a work". The idea is that we can have at least two distinct and interconnected level when we talk about a book (or any other kind of creative thing): one is recognising the book independently from its actual content (the English translation, the first edition in French, etc.), e.g. when I discuss about Harry Potter and the Chamber of Secret with my English and German friends, independently from the fact I've read it in Italian and they have read it in their own languages. In this case, we are all speaking about the same thing (which we have identified here as a work-item). Another level we can use is to refer to a particular edition of that work: the first Italian translation, the very first edition published, etc. This level (i.e. the edition-item) is clearly interlinked the related work-item, but they actually describe two different perspectives of something that, cognitively, is erroneously perceived as the same thing. Essepuntato (talk) 05:41, 30 September 2016 (UTC)

@billinghurst: Do a choice: you can't say we create false structure if at the same time you accept the concept of work. A work item is not a real thing, it is just a concept used to link some formats. That's a wrong structure in your trail approach based only on the real sequence of format apparition. And what you neglect by your approach is the possibility to avoid duplicated data. In a trail approach, as you don't know the length and format of the items, you have to use a recursive programm to extract information like original language, original publication or author, this leads to the addition of these data to all items of the trail to avoid to have to go the first item of the trail. With our approaches we aims to model the sequence of editions (nothing prevent you to link a translation to the edition in original language using a "translated from" property) and to center some common data in a conceptual item offering a way to connect all items based~on the work together. Snipre (talk) 05:58, 1 September 2016 (UTC)

Respectively saying this but that simply sounds nonsensical to me. I am saying that we have an origin, and it is what it is. This attempt to classify it to a variety of a work seems to me to be false and too open to interpretation and reclassification. As soon as you start to get surreal this all just becomes too hard and the integrity of a classification just fails due to interpretation. Works have a linearity, and sometimes connectivity. here the project is Wikiproject books, so I suggest that books are the focus, and we work out how these books interrelate to editions which sit beside not subsidiary to books.  — billinghurst sDrewth 13:22, 1 September 2016 (UTC) [Please note that my "you" is the collective/plural, not the singularly. My commentary is about the discussion]

@billinghurst: IMHO, there is one thing that distinguishes a work by another: the author! When we have another primary authorial responsibility, we have another work even if "the origin" is the same...but I could be agree focus on literary works first of all.Nonoranonqui (talk) 13:44, 13 October 2016 (UTC)

discussion at main Project Chat

There is a discussion regarding the use of via (P2825) regarding books available from sources such as the Internet Archive or googlebooks.

Any thoughts are most welcome. Robevans123 (talk) 12:57, 10 September 2016 (UTC)

Struggling with "based on" being an equivalent of "adaptation from"

Hey All, could use some feedback on the use of Property:P144 (Based on) to represent the concept of adaptation in creative works: Property_talk:P144#Splitting_.22adapted_from.22_from_this_property, Sadads (talk) 17:20, 10 September 2016 (UTC)

Hi Sadads: my fear is that a "adaptation from" property as you suggest would be very specific, academic-wise, and that nobody would understand the difference with "based on" or even (maybe) "edition of". I don't know anything about the academic literature about adaptations but it seems to me that it's very detailed. If people don't understand it you would end up just adding noise to Wikidata... But maybe I'm too pessimistic. Aubrey (talk) 08:33, 13 September 2016 (UTC)

Speculative fiction and fantastic

Hello WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.! I have two things i want to discuss:

1. Problem with items speculative/fantastic fiction (Q5967378) and speculative/fantastic fiction (Q2387832). English article from speculative/fantastic fiction (Q5967378) best corresponds to term "fantastic" from a lot of articles in speculative/fantastic fiction (Q2387832). w:en:Fantastique from speculative/fantastic fiction (Q2387832) pretty much describes history of french speculative fiction, and w:ru:speculative fiction, w:uk:спекулятивна белетристика, w:bg:Спекулативна фантастика and maybe more from speculative/fantastic fiction (Q5967378) describe English term for fantastic, how it came to be, and that today it is general term for fantastic and sends for further reading into corresponding article from speculative/fantastic fiction (Q2387832). Basically speculative fiction and fantastic ("фантастика") mean the same and should be in one item on wikidata. As for fantastique in English wiki, it can be linked to other articles in speculative/fantastic fiction (Q5967378) and renamed to something like "history of speculative fiction" in all wikis to avoid confusion. Of course i can say this only for few languages, but both of those items can't be used to describe genre (P136), or used as subclass of (P279) in other fantastic genres.
2. Suggestion to use items like short story (Q49084) and novel (Q8261) in property instance of (P31) not in genre (P136). It's related to use of wikipedia templates, which will display something like "genre: science fiction, short story and horror fiction", if those items used in genre (P136). I know that short story (Q49084) and novel (Q8261) usually described like genres, but they stand out in templates under genre field, since they basically describe size of work not themes.

--Similartothissimilartothat 23:58, 27 September 2016 (UTC)

Hi, I'm not really competent on the first question, but on the second one, I agree completely : on fr.wikisource, we identify our texts as short story (Q49084) and novel (Q8261), disconsidering genre (P136), as a simple indication of short fiction and long fiction. On wikisource items I used them in instance of (P31) too... --Hsarrazin (talk) 08:06, 28 September 2016 (UTC)

Hi Similartothissimilartothat,

I feel like that these two questions maybe a linguistic and cultural ones.

For your first question, clearly there is some things wrong like

⟨ speculative/fantastic fiction (Q2387832)    ⟩ topic's main category (P910) ⟨ Category:Speculative fiction (Q7005415)    ⟩

but globally it seems good to me. I think We need to item and « Speculative fiction » should be a superclass for speculative/fantastic fiction (Q2387832), science fiction (Q24925), fantasy (Q132311) (but right now, the two last ones are subclass of (P279) of both speculative/fantastic fiction (Q5967378) and speculative/fantastic fiction (Q2387832) which seems wrong to me).

To simplify and (hopefully) to be clearer, is see things like that : science fiction (Q24925) is for books with robots and aliens in a futuristic touch, fantasy (Q132311) with elves and dragons in a magic or medieval touch, speculative/fantastic fiction (Q2387832) with supernatural beingsphenomena and speculative/fantastic fiction (Q5967378) is for bookshelves where we put these three types of books.

For the second question, it should'nt be used in instance of (P31) as it clearly belong to genre (P136) to me and to many specialists. Hsarrazin are you sure ? fr:s:Catégorie:Romans and fr:s:Catégorie:Nouvelles are clearly under fr:s:Catégorie:Genres littéraires !

Cdlt, VIGNERON (talk) 08:34, 28 September 2016 (UTC)

Hi VIGNERON, these problems might be ones of language and translation indeed.

I agree with you about "Speculative fiction" as superclass for speculative/fantastic fiction (Q2387832), science fiction (Q24925), fantasy (Q132311). :)

On the short story (Q49084) / novel (Q8261) are clearly forms of texts (short fiction, long fiction) - not genre which is something else, like romantic, fantastic, realistic, etc.

since book (Q571) clearly cannot be used on those when they are considered separately from the physical unit (published in periodicals for example), what instance of (P31) would you suggest ? text (Q234460) is much too wide, and incorporates items that are not "works".

seems to me that we already had this discussion 2 years ago, and the then chosen item was merged in something else, which disturbed me a lot :( -- maybe I'm not totally up-to-date with the project ? sorry if I re-introduce long solved questions... :) --Hsarrazin (talk) 09:46, 28 September 2016 (UTC)

The problem seems clearly linked to langage, in French fr:Roman (littérature) is clearly a genre (with a prefered - but not mandatory - form ; but the form came from the genre, not the other way round). In other languages, it's less clear, see en:Novel in English (which cover two words of novel and romance, where French has only one word...).

Maybe a solution is to use more precise sub-items, like science fiction novel (Q12132683), crime novel (Q208505) (I think we all agree this is a genre).

For the use of instance of (P31), that's an entire different and unsolved problem (discussion are still going on above, please don't mix and scatter discussion as this is complicated enough ; let's focus on the 'genre' here).

Cdlt, VIGNERON (talk) 10:23, 28 September 2016 (UTC)

Use of more precise subitems could be a solution, but only if there will be a rule in case of literature work that has two genres. For example lets assume existence of work that is both fantasy and detective. If just more precise items are used, wikipedia template will show: "genre: fantasy novel and detective novel", which again sounds bad. But if genre (P136) will be edited to show first item "fantasy" and second item "detective novel", than template will show "genre: fantasy and detective novel" which is much better.--Similartothissimilartothat 11:29, 28 September 2016 (UTC)

Similartothissimilartothat very true, but you're pointing an unusual case, most of the cases will be (I hope) more simple. And in your case, a simple solution is a « fantasy-detective novel » sub-sub-item (and yes this is a thing, there is even a sub-sub-sub-item for occult detective fiction (Q10728648) !). Cdlt, VIGNERON (talk) 08:49, 29 September 2016 (UTC)

Hi VIGNERON. It is indeed a translation problem. Term "фантастика" in slavic languages encapsulates science fiction (Q24925), fantasy (Q132311), horror fiction (Q16575965), and most fiction genres. Which I can easy fix by moving articles with that term from speculative/fantastic fiction (Q2387832) to speculative/fantastic fiction (Q5967378) (also I would prefer to move "speculative fiction" into speculative/fantastic fiction (Q2387832) since it used more often as genre on wikidata ). But then there would be a problem if "fantastique" will be used to describe genres, since, as far as I know, our language doesn't have corresponding term (most will connect it to "фантастика", but as i said it encapsulates other fiction genres the same as speculative fiction today). As for meaning of "fantastique" i didn't heard of genre about supernatural beings. If source of supernatural state of this beings explained in some sciency way it would be put in science fiction genre, and if leaved to reader imagination or pinned on magic, it would be under fantasy category.--Similartothissimilartothat 10:55, 28 September 2016 (UTC)

Hi Similartothissimilartothat. Are you sure that « encapsulates » is the right term? In French, "fantastique" is at the crossroads of science fiction (Q24925), fantasy (Q132311), horror fiction (Q16575965) but it didn't really « encapsulates » them (and en:Fantastique agree a bitas it says « overlaps » ; this article has a definition very close from the French one : « What is distinctive about the fantastique is the intrusion of supernatural phenomena into an otherwise realist narrative. »). As I understand what you say and from rough Google translation of articles, I think that in slavic languages, фантастика looks indeed closer to speculative fiction but I don't see how we can deal with this... (except maybe with mirabilia (Q3306693) which is sometimes mixed and between tales and fantastique in French, sadly I didn't find equivalent in slavic so I don't know if it can help or not...) Cdlt, VIGNERON (talk) 12:02, 28 September 2016 (UTC)

I must repeat that the "genre" property is cumbersome and should be replaced by instance of (P31). All of this is a matter of work classification. Let's classify work over clear criterias and not fight over terminology issue :) This class could include the type of narrative and the length of the text, for example, and in a next step we could classify the classes themselves. author  TomT0m / talk page 11:12, 28 September 2016 (UTC)

Well it's maybe cumbersome but I don't quite see how to do with instance of (P31)... For instance, how can infobox easily get the genre of a book? More complexe, constraints checking: how can we check that every items about books has a value that is a genre? Cdlt, VIGNERON (talk) 12:02, 28 September 2016 (UTC)

@VIGNERON: Constraint checkink would be quite easy. As genre-instances are classified in the class "genre" - whatever the definition is, we can even make different genre classes depending on the culture - it's enough to check that any literary work is also an instance of a class that is a genre. A contraint template to do "if ... the book is a literary work then it is instance of a genre would be {{IfThenConstraint}}. The "the book is a literary work" is straightforward and doable with {{Query instances|thebook|literary work}} and the "it is instance of something that is a genre" is doable with something like ?thebook wdt:P31/wdt:P279*/wdt:P31 wd:genreitem. author  TomT0m / talk page 13:59, 28 September 2016 (UTC)

Hi, I think Vigneron is right. There are clearly 3 genres into "Speculative fiction": while science fiction (Q24925) and fantasy (Q132311) are well known, good examples of speculative/fantastic fiction (Q2387832) are vampire stories and Charmed (Q162371). Regards, Yann (talk) 12:08, 28 September 2016 (UTC)

Yann Vigneron isn`t right. In slavic, baltic and some east languages en:The Day of the Triffids, en:Mythago Wood, en:It (novel), en:The Man in the High Castle, - This is All"uk:фантастика".--Yasnodark (talk) 14:50, 28 September 2016 (UTC)

Yasnodark : What you say is not a refutation of what Vigneron and I say. There may be fictions with elements from more than one genre, but it doesn't change the clear division of 3 main subgenres. You can even explicitly create a story for the purpose: include a dragon in a future world with aliens and vampires, you are done... Literature is not defined in advance. Hopefully... ;oD Regards, Yann (talk) 21:24, 28 September 2016 (UTC)

Similartothissimilartothatfull rights terminology in all Slavic languages ​​and other languages ​​of the former USSR "Фантастика" is fully consistent with the term "Speculative fiction". And combines science fiction, fantasy, horror, and an alternative history, I suggest you use the same perspective "Fantastique" term - it is convenient and practical. There will be two-fold meaning. And VIGNERON the sub-genre that you & some otters used to be called Fantastique in French, such as call Supernatural Fantastique. "Fantastique" term use a set or combination of science fiction, fantasy, horror, and an alternative history.

I think the solution would be: in a group binding - the basic articles for the language in the other - explanatory concepts and terminology duplicate terms. So that the reader of Wikipedia in different languages, going for interwiki got to truly appropriate within the meaning of the article, not the limited (in their objectives) conceptual and terminological article, which tells about the unimproved in this language term duplicate, common in other languages. Incidentally shortened form of interwiki for ordinary readers almost makes it impossible opportunity to move to other sections for them. In english en:Speculative fiction is basic term, in slavic, baltic and east languages - "Фантастика": "uk:фантастика", "ru:фантастика", "bg:Фантастика", "pl:Fantastyka", "uz:Fantastika", "lt:Fantastika", "et:Ulme", "be:Фантастыка", "sh:Fantastika", "kk:Фантастика", "sv:Fantastique" or "lv:Fantastika". In english dublicate term is en:Fantastique, in slavic languages - "ru:Speculative fiction" or "uk:Спекулятивна белетристика" is duplicate term. Therefore it is necessary to link the basic terms for the main and secondary to secondary.

be-nt-all from Russian wiki have the same opinion about this question:Speculative fiction=Фантастика ru:Обсуждение проекта:Вымысел. Ulv80 from Polish wiki - I think too, judging by his editing.--Yasnodark (talk) 14:50, 28 September 2016 (UTC)

Classification of genre

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

The discussion above is very interresting and I like to open a new subject to go further. Before changing anything, I think we should look 1. at the big picture (all genres of fiction, not just fantastique/speculative) 2. look at the sources.

For the first point, we should start by doing a diagnostic of the current situation. I began by building a graph of the subclass of (P279) of fiction (Q8253) : [1]. I was surprised to see that the tree is very little, we have a lot a classification to do (and while some may be complexe, I think that most of them will be easy). Here is a list of subgenre without subclass of (P279) :

SELECT ?item ?itemLabel WHERE {
	?item wdt:P31 wd:Q223393 .
	minus { ?item wdt:P279 [] } .
	SERVICE wikibase:label { bd:serviceParam wikibase:language "fr" }
}

Try it!

For the second point, classification shouldn't be base on what we think but on what the source say. It's a bit more difficult to find sources but in the long run, it will be more reliable and better for everyone. It will be especially useful for languages and cultural differences. I hoped to use the Dewey Decimal Classification (Q48460) whi is universal but it's not very good at fiction and I only have the French version (I know it has been translated in many languages though, so it still could help but it won't solve everyhting).

What do you think?

Cdlt, VIGNERON (talk) 08:45, 29 September 2016 (UTC)

I would not count classification of literary genre (Q223393) instances (and its subtypes) as major part of this project. Any categorization will be too fuzzy and subjective to be of much use. Surely literary studies (Q208217) and library science (Q199655) can give some input but to start with it's enough to distinguish fiction and non-fiction. -- JakobVoss (talk) 10:23, 29 September 2016 (UTC)

Not a major part but a part nonetheless and an useful one might I say. It could be subjective, that's why I ask sources to rely one ; it will still be subjective but with multiple and referenced points of view, which is probably the less worst we can manage to achieve. Cdlt, VIGNERON (talk) 13:16, 29 September 2016 (UTC)

There are only 2 overgroups of genres of literature or directions:

• Fiction and Non-fiction

There are only 3 groups genres of fiction or fiction directions:

• • 1.en:Speculative fiction - en:Science fiction (Hard science fiction with subgenres & soft science fiction, for example, utopia and dystopia), en:fantasy & en:fairy tales, apocalyptic & postapocalyptic fiction, en:horror fiction, and an en:alternative history.

• • • 2.en:Literary realism - realistic genres:
    • mainstream genres (philosophical fiction, psychological fiction, historical fiction, social fiction, satirical fiction & child realistic fiction and others) and
    • pulp genres (detective, mystery & crime fiction; love, romance & erotical fiction; vesterne, triller & action fiction and others)
    • &border genres between en:mainstream & en:pulp fiction (humorous fiction, mariner & adventure fiction and others)

and

• • 3.Border genres between en:Speculative fiction & en:Literary realism - such as poetry, magic realism, surrealism, a quasi-historical novel, сryptohistory ets.

There are groups genres of non-fiction or non-fiction directions:

• • 1.scientific non-fiction
  • 2.popular scientific non-fiction
  • 3.biographical and autobiographical non-fiction
  • 4.journalistic non-fiction and others.--Yasnodark (talk) 12:53, 29 September 2016 (UTC)

@Yasnodark: only is very authoritative, can you provide references? I've borrowed and I'm currently reading books on the subjects (Le genre littéraire, Marielle Macé, 2013 and Qu'est-ce qu'un genre littéraire ? Jean-Marie Schaeffer, 1989) ; I just began but they're saying that there is a lot more of genres with subgenres and mixed-genres, like a spectrum of genre (and books are not categorised only by genre, but with others criterias as tone (Q3423650) or style). Cdlt, VIGNERON (talk) 13:20, 29 September 2016 (UTC)

I agree with @VIGNERON, JakobVoss: genres are a can of worms (talking with my librarian hat now), but of course it's helpful for the user if we can provide a genre with a source. Publishers and libraries classify books per genre, so it's doable. Aubrey (talk) 10:11, 30 September 2016 (UTC)

Handle fictious identity

Please see this. We should have guidelines on how to handle pseudonyms, and this is not just for books. author  TomT0m / talk page 10:45, 20 November 2016 (UTC)

I would have thought that pseudonyms (to known persons) would all be handled on the author page, and we simply link to the known author. If there is no known author (pseudonym or not) then we would be using the author = "unknown value", knowing that they can be updated if the author is ever known.

If there is more required than that then it is probably worth summarising what are the known issues for resolution here (noting that your linked discussion will soon disappear to the archives and be of little value, so could you please put in a permalink.)  — billinghurst sDrewth 12:57, 20 November 2016 (UTC)

Permalink will come in the archive when the time is right ... But please discuss there or we won't have a central point to discuss. I'll raise your concern there. (It's the same for discussion on this page, they'll be archive soon). author  TomT0m / talk page 14:06, 20 November 2016 (UTC)

Universal Short Title Catalogue (USTC)

Did someone already check whether http://www.ustc.ac.uk/ is of interest for us? --Nemo 13:34, 2 December 2016 (UTC)

multi-work/composite editions

Dear WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead., how would you structure an entity that is a composite edition or translation of (P629) several work (Q386724): for example, an edition of the 3 greatest works of an author major works? Should it simply have several values for edition or translation of (P629), or should it have for edition or translation of (P629) an ad-hoc created composite work entity? In this last case, what would be the relation between this composite work and the original works?

The option that looks the less bad is the first I guess - setting several edition or translation of (P629) - but it could then be useful to have a specific instance of (P31) value instead of version, edition or translation (Q3331189), something like instance of composite edition. Do we have such a thing?

Some context: I'm not aware of such a need within Wikidata at the moment, but this is a problem I have on inventaire, where I'm restructuring the database as an open and contributive database on books, extending Wikidata properties and entities. The expected result is CC0 data on books ready to be imported in Wikidata, thus the desire to have this discussion ahead.

Thanks for any clue -- Maxlath (talk) 12:58, 8 December 2016 (UTC)

@Maxlath: Examples would help. If a work has been published that has parts that have been separately published, I would have handled the work to be "book" which is its own entity, and it has parts. The reproduced parts would separately be itemised as "editions of" linking back to their individual "literary work"; and the literary work would have "edition(s)" links to the parts. I would have thought examples of this are things like an author's compiled works which were individually published that can later be reissued in a variety of forms.  — billinghurst sDrewth 07:39, 9 December 2016 (UTC)

@billinghurst: Example: the ISBN 9782878581126 refers to a joint edition of The Chalk Circle Man (Q3203603) and L'Homme à l'envers (Q3203802). But as that's a one-shot pack edition, I wouldn't find it very elegant to create a new dedicated work, or we will end up with authors having written 10 book (Q571) but ending up with a list of 100 attributed works, just because publishers recomposed their work in all the possible ways. -- Maxlath (talk) 11:47, 9 December 2016 (UTC)

I think edition or translation of (P629) should pretty much always just have one value. For editions that contain multiple separately-published works we should use has part(s) (P527), and have an item for the work as well. For example, the following three works and their editions:

1. Work: Moll Flanders (Q612523) — instance of (P31)book (Q571) part of (P361)Moll Flanders and Roxana (Q29053398) (links to Wikipedia)
   • Edition: Moll Flanders — instance of (P31)version, edition or translation (Q3331189) edition or translation of (P629)Moll Flanders (Q612523)
2. Work: Roxana: The Fortunate Mistress (Q3825485) — instance of (P31)book (Q571) part of (P361)Moll Flanders and Roxana (Q29053398) (links to Wikipedia)
   • Edition: Roxana (Q19102887) — instance of (P31)version, edition or translation (Q3331189) edition or translation of (P629)Roxana (Q19102887) (links to Wikisource)
3. Work: Moll Flanders and Roxana (Q29053405) — instance of (P31)book (Q571) has part(s) (P527)Moll Flanders (Q612523) has part(s) (P527)Roxana (Q19102887)
   • Edition: Moll Flanders and Roxana (Q29053398) — instance of (P31)version, edition or translation (Q3331189) edition or translation of (P629)Moll Flanders and Roxana (Q29053405) (links to Wikisource)

(Sorry, I gave up sorting out the edition of the first one; but hopefully you get the idea.)

This question has come up recently at Wikisource, where I'm attempting to sort out a system of displaying sitelinks based on edition or translation of (P629). It seems that there's some uncertainty about the correct structure on Wikidata though, so I've come to ask here. (Pinging @billinghurst, EncycloPetey, Hrishikes.)

@Maxlath: "we will end up with authors having written 10 book (Q571) but ending up with a list of 100 attributed works, just because publishers recomposed their work in all the possible ways." I agree, it could be a bit over the top, but this is how library catalogues do it, and I'm not sure how we can represent the data in any other way.

—Sam Wilson 05:28, 31 March 2017 (UTC)

My point at Wikisource was like this: Wikipedia contains many articles on books and stories. For each article on a mythological story or a fairy tale, there may be multiple works in Wikisource (same WS, or multiple WSes), where that story is included, in works translated/edited/compiled/retold by different authors. There are multiple works on Indian mythology/fairy tale in English Wikisource, where the same story is sometimes found in different works. When such a story is transcluded, it looks better if the concerned Wikipedia article is linked. I am currently doing it manually. Hrishikes (talk) 06:13, 31 March 2017 (UTC)

@Samwilson: re: "I think edition or translation of (P629) should pretty much always just have one value". You haven't explained how we make that happen, though, and haven't even given an example that addresses the issue. Here's a specific instance that could address the issue: Euripides wrote a satyr play called The Cyclops. Shelley did a famous translation of this work in English. There are two editions (different text) of Shelley's translation, which I will call "A" and "B". So, does what value for edition or translation of (P629) do we give "A"? It is an edition of the Shelley translation, and it is a translation of Euripides' play. So it is an edition/translation of two separate things. This problem is the direct result of having a single property for both "edition of" and "translation of". --EncycloPetey (talk) 14:08, 31 March 2017 (UTC)

No, I was just trying to clarify the "composite editions" question that started this thread. But yours is a good example. I'd say that both of Shelley's translations are editions of the work Cyclops (Q1206121) (with original language of film or TV show (P364) set appropriately). I think the reasoning for having 'edition or translation' is that a translation is an edition, and must always be. Even if it were a translation of a translation, it would still be a translation of the original work (with based on (P144) set as well probably). The only other thing that could be done is to also have separate items for the work (play), expressions (whatever Shelley translated from), and manifestations: an item for the play, one each for the different manuscripts that survive for it, and then as many editions of these as are required (including translations). Sam Wilson 04:44, 7 April 2017 (UTC)

I disagree that a translation is an edition, and do not understand what you are proposing as a solution for situations like that of Shelley's translation. This is not an isolated situation either. Elizabeth Barrett Browning made two translations of Prometheus Bound (original by Aeschylus), and there are several printed editions of the Thoreau translation of the same play. --EncycloPetey (talk) 08:54, 7 April 2017 (UTC)

Lua error: too many expensive function calls.

@Micru: The obverse to this page is showing errors Lua error: too many expensive function calls. Are you able to do something to assist?  — billinghurst sDrewth 00:20, 12 December 2016 (UTC)

As a temporary measure I have rem'd component containing template:authority control properties as it is of lesser importance. I actually question whether it is needed on this page, rather than a link to a separate page.  — billinghurst sDrewth 00:47, 12 December 2016 (UTC)

A possible solution could be to split the content in several subpages.--Micru (talk) 09:02, 12 December 2016 (UTC)

Addressing issues of audio files

Hi. I believe that we need to have some direction to "audiofiles" in the edition space. We should be giving direction to users how we wish to identify and link audio copies of existing editions. What factors should we be adding? what qualifiers? how do we link to Commons-hosted recordings? and especially make such information easily usable at the wikis if there is a hosted edition at a Wikisource.  — billinghurst sDrewth 01:35, 20 December 2016 (UTC)

Doubts about main work and editions

Hi everyone!

I have one doubt about the way to add books in Wikidata. If I want to add a book with only one edition, have I got to add another item with the edition or it could be only the same mixing the work and edition properties? For example, Patrimonio arquitectónico y artístico de la Universidad de La Laguna has been published one time so I decide to use edition properties in that.

Thanks in advance!

Regards, Ivanhercaz | Discusión 02:45, 28 December 2016 (UTC)

@Ivanhercaz: Your simplification is a just a nightmare for data extraction: we will need 2 different codes to extract data with your contribution. One code for books where there are different items for work and editions and another code for book where work and edition are mixed in the same item. So please if you want to do the minimal work do it correctly: please specify that:

• Patrimonio arquitectónico y artístico de la Universidad de La Laguna (Q27653258) is an instance of book AND an instance of version, edition or translation (Q3331189)
• Patrimonio arquitectónico y artístico de la Universidad de La Laguna (Q27653258) is a edition or translation of (P629) of Patrimonio arquitectónico y artístico de la Universidad de La Laguna (Q27653258)
• has edition or translation (P747) number of Patrimonio arquitectónico y artístico de la Universidad de La Laguna (Q27653258) is 1.

Snipre (talk) 13:59, 28 December 2016 (UTC)

@Snipre: It is not about to do the minimal work, it was a doubt about which is the correct way. Sorry if the question was misunderstood. I ask because I want to make my work correctly and not «nightmare for data extraction». So I am going to make another item for the edition when I have time (I am going to try that night) .

I have doubts with has edition or translation (P747) because you have said that it have to be 1, but in the Wikidata property example of has edition or translation (P747) is a item, Die Götter Griechenlandes (Q1147941), with its respective edition property that shows a Portuguese and a Russian editions.

Another doubt, when I will create the item for the edition, should I add the number of edition between brackets in name of the item? I mean something like Patrimonio arquitectónico y artístico de la Universidad de La Laguna (1 ed.).

I am going to improve the item! Thank you Snipre, if you (or someone) have (has) another advice I will be grateful, because I think that add books is a delicate task, so I want to make it as best as possible.

Regards, Ivanhercaz | Discusión 17:53, 28 December 2016 (UTC)

P. S. I am going to improve Guinea Ecuatorial, país joven: testimonios políticos (Q28062489) too with your advice because I create the item only for the work and not for the edition.

Hi again! Snipre, I have created the item for the edition of Patrimonio arquitectónico y artístico de la Universidad de La Laguna (Q27653258): Patrimonio arquitectónico y artístico de la Universidad de La Laguna (1 ed.) (Q28101541). Now I am going to improve Guinea Ecuatorial, país joven: testimonios políticos (Q28062489), but I would like to know if I create finely the item for the edition. Have I forgot something or made something wrong? Thanks in advance.

Regards, Ivanhercaz | Discusión 23:21, 1 January 2017 (UTC)

P. S. And there is Guinea Ecuatorial, país joven: testimonios políticos (Q28062489) and the item of its first edition, Guinea Ecuatorial, país joven: testimonios políticos (1 ed.) (Q28102071).

@Ivanhercaz: interestingly I have gone away from the idea of a "book" and thinking of the top level as the literary work/idea/concept, and then the editions being the manifestations of the concept. For me this is especially relevant where I have been reproducing 19th century speeches, lectures. etc. so they are the base where I start, and the editions will come from that. On the general point of our master / slave representations, I will admit to being lazy on many occasions and only doing the "edition", so we do need a ready tool that allows easy creation of the edition from the master, and the creation of the master from the edition.  — billinghurst sDrewth 23:26, 3 January 2017 (UTC)