Open tombaker opened 6 years ago
I think it's confusing that it isn't a complete citation, although I think i understand why it isn't. A complete citation would include the author and title and dates, whereas this assumes that those are coded as dcterms and would be redundant here. However, the definition is: "A bibliographic reference for the resource." To me, this is a bibliographic reference:
Vannevar Bush. “As We May Think.” The Atlantic, July 1945, 176 (1): 101–108.
Using the example above, dct:bibliographicCitation would be(?):
The Atlantic, 176 (1): 101–108.
If this is what is intended, then the comment needs to say that this is the remainder of the bibliographic information that is not already coded in other fields, such as journal titles and enumeration. However, to call that a "bibliographic citation" is pretty misleading. Note also that for books, in most cases you would not have anything to put in dct:bibliographicCitation since it is all covered by other fields.
The upshot is that I think the example should be a full citation, or the comment needs to change to explain that this isn't a full citation.
I find this one to be quite puzzling.
Agree with @kcoyle , either insist on a full citation (with corresponding example) or else explain what goes in this field and what doesn't.
It is not entirely clear to me what services dct:bibliographicCitation needs to support. If we decide that it should be possible to extract this one data element from DC record and create a bibliographic citation from it, full citation (whatever that means) is needed. But the current definition may be understood in other ways as well.
If we decide to require a full citation, we have to choose one or more commonly used citation styles and provide examples based on them (including punctuation). ISO 690 (ISO standard for bibliographic citations and references) does not mandate any given style, but gives examples of different citation styles. There are currently about 8000 of them.
I have no strong opinion either way (full citation / just sufficient detail needed for the identification of the cited resource), but it is easier to use the data in dct:bibliographicCitation if there is no need to use metadata from other elements of the DC record to create a full citation. And "sufficient bibliographic detail to identify the cited resource unambiguously" may mean just the identifier of the resource, and since the identifier is provided already in dct:identifier, I see no need to repeat it.
I have restored the original comment to the ISO draft for now, and removed the example which was admittedly confusing. But I do find "Recommended practice is to include sufficient bibliographic detail to identify the resource as unambiguously as possible" just a complicated way of saying "Recommended practice is to include a full citation", and would prefer this shorter formulation.
I want to support the case for a full bibliographic citation, because it is particularly useful in linked data appllications as a (link) label for a publication. If you have to create such a label on the fly through a SPARQL query, you end up with lots of OPTIONAL clauses, and at the same time with lots of GROUP BY / GROUP_CONCAT contructs, because few if any elements are guaranteed to exist in a given dataset, and almost every one can occur more than once. Both constructs are expensive in terms of query complexity and of processing time.
Reacting to several aspects here:
-1 for "Recommended practice is to include a full citation" only. This is quite circular, and I think that mentioning the need to 'identify' (which I understand here to also include a bit a motivation for 'find') is important. It feels a bit moot in a context where the subject of the statement (and the metadata this subject is attached to by other statements) already hopefully describes the resource in more detail than needed, as noted a couple of times in this thread. But I can understand the need for having a recommended citation for the resource, for example so that (1) it can be included in old-school citation lists or (2) (linked data gods protexts us from this!) the identifier in the Linked Data resource or the dct:identifier would break and the citation would be the easiest way to find again the resource using general search engines or scholarly repositories.
Hence +1 for @kcoyle 's proposal a full citation. The recommendation should that the literal goes as far as possible in what it enables. Otherwise the motivation for dct:bibliographicCitation is even less clear.
+1 for having a note that refers to come citation styles (as @juhahakala suggested)
The comment should not include something like "to identify the cited resource unambiguously". To me 'cited resource' could refer to a resource cited by the described resource (the subject of the dct:bibliographicCitation statement). We should avoid such ambiguity.
See note_bibliographicCitation.md.
Add comment for http://purl.org/dc/terms/bibliographicCitation:
Note: