Wikidata:Property proposal/ISBN publisher code

From Wikidata
Jump to navigation Jump to search

ISBN publisher prefix[edit]

Originally proposed at Wikidata:Property proposal/Creative work

Descriptionpart of an ISBN specific to a publisher
Data typeExternal identifier
Domainpublisher (Q2085381)
Allowed values97(8|9)-[0-9]{1,5}-[0-9]{2,7}
ExampleScottish Academic Press (Q7437604) → 978-0-7073
Sourcehttps://en.wikipedia.org/wiki/List_of_group-0_ISBN_publisher_codes
Planned usehaving this property would allow to find the publisher (P123) of a given version, edition or translation (Q3331189) easier
See alsoISBN-13 (P212), ISBN-10 (P957)
Motivation

Having this property would allow to find the publisher (P123) of a given version, edition or translation (Q3331189) more easily.

The remaining issue is that an editor identifier is only unique to a given language block. For example: 0-14-043531-X

  • 0 = English language
  • 14 = Publisher code
  • 043531 = Item number
  • X = checksum

(example borrowed to OpenLibrary)

Here 14 identifies Penguin Books (Q1336200) (source), but it is specific to the 0 (English language) block; in the 2 (French language) block, 14 is attributed to L'Harmattan (Q3579383) (source). I see two ways to solve this issue:

  • create an other property specifying to which language block an editor identifier is bound (but just as some publishers got attributed several codes, some might be spread upon several language blocks?)
  • turn this property into "ISBN publisher prefix" integrating also the language block: in the previous example, instead of just 14, that would be 0-14

WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. Zorglub27 (talk) 16:34, 15 July 2016 (UTC)[reply]

Edit 1: in the case we go for the latter, we might even need to include the GS1 prefix (currently limited to 978 and 979 for books), which in the above example would give 978-0-14.

Edit 2: Following the discussion, I updated the proposal to include the GS1 prefix and group identifier: this resulted in changes in the property title (from "ISBN publisher identifier" to "ISBN publisher prefix") and in the "allowed values" regular expression (from /[0-9]{2,7}/ to /97(8|9)-[0-9]{1,5}-[0-9]{2,7}/ ). @John Vandenberg, VIGNERON, Thryduulf, Accurimbono, T.seppelt, Skim: could you review this updated version and eventually confirm your support? Zorglub27 (talk) 11:09, 18 July 2016 (UTC)[reply]

Discussion

 Comment not sure if we want this into Wikidata... It is definitely interesting, but maybe our of scope? Aubrey (talk) 16:42, 15 July 2016 (UTC)[reply]

  • This is really useful info to have in the database. I've been intending to write this up, but you have done a wonderful explanation. Thanks. Storing as a prefix (i.e. 0-14) is my preference, as it allows data constraints (i.e. no dups) to be created and enforced. Storing ISBN country prefixes somewhere can be useful, with the range they are granted, but that dataset changes periodically, and IMO only invalid conclusions can be made by using this country data, as it only exists to delegate a namespace to an ISBN agency. John Vandenberg (talk) 23:17, 15 July 2016 (UTC)[reply]
  •  Support a bit redundant with ISBN full code but could be very useful to check the publisher ; for French publisher, there is a database here : [1]. VIGNERON (talk) 09:27, 16 July 2016 (UTC)[reply]
    I don't think it would be redundant, given that without it we have no mean to make the connection from a given ISBN to the associated publisher without manually searching through data scattered all around the web, or even missing data Zorglub27 (talk) 10:30, 16 July 2016 (UTC)[reply]
    @Zorglub27: well, if you have ISBN = 0-86140-324-X on the edition book and 86140 to the linked publisher item, it is technicaly a bit redundant, you could have infered automatically that 86140 is the number for the editor ; it doesn't mean that it is a bad thing, on the contrary, redundancy is very useful for checking ;) Cdlt, VIGNERON (talk) 09:21, 17 July 2016 (UTC)[reply]
  •  Support great --Accurimbono (talk) 11:05, 16 July 2016 (UTC)[reply]
  •  Support including the language code so the identifier is unique. Thryduulf (talk) 12:16, 16 July 2016 (UTC)[reply]
    @Thryduulf: yes, and in this logic, we might even need to include the GS1 prefix (currently limited to 978 and 979 for books), which in the above example would give 978-0-14. Zorglub27 (talk) 16:47, 16 July 2016 (UTC)[reply]
    @Thryduulf, Zorglub27: interesting idea, great for the « lang » code (not really for lang, more the « registration » code, but still a good idea) but I'm not so sure if the GS1 prefix is really useful... Cdlt, VIGNERON (talk) 09:21, 17 July 2016 (UTC)[reply]
    @VIGNERON: I think it would be "future-proof" to include those GS1 prefix, given that registration code defer between 978 and 979: for example, without it, Glose would be given the value 10-01, with no indication that it actually belong to the 979 domain. We could just infer it from the fact that apparently the 928-10 domain isn't used, but that's not really convenient. I couldn't find an example of direct collision but I guess those will come sooner or later as it's the whole point of using a new GS1 prefix in addition to 978. Zorglub27 (talk) 10:20, 17 July 2016 (UTC)[reply]
  •  Support -- T.seppelt (talk) 14:52, 17 July 2016 (UTC)[reply]
  •  Support with prefix. -- Skim (talk) 08:22, 18 July 2016 (UTC)[reply]
  •  Support. Note that it will probably need to specify that this is for long ISBN (ISBN-13) numbers (the regular expression does not include ISBN-10). — Finn Årup Nielsen (fnielsen) (talk) 13:12, 8 August 2016 (UTC)[reply]

Post-Edit 2 discussion:

@Lymantria: indeed! Corrected. -- Zorglub27 (talk) 10:00, 6 August 2016 (UTC)[reply]

@Zorglub27, JakobVoss, Thryduulf, John Vandenberg, Skim, VIGNERON:✓ Done Now ISBN publisher prefix (P3035). --Lymantria (talk) 06:57, 7 August 2016 (UTC)[reply]