> Just to bring things back to where (I think) we started. I think people > are > talking about three separate things here: > > * URIs for bibliographic works (which, as Karen pointed out, are > missing > some crucial bits of info like page numbers) "URIs for bibliographic works": I assume you mean "works" in the sense of frbr:Group1 or possibly owl:Thing. (As an aside, does code4lib ever talk about OWL?) > * Actual text representations of a citation in a variety of > thick-book-specified formats (e.g., ALA, MLA) Consider the URI patterns I suggested. > * The cites/cited-by graph for everything everywhere. If "cites" and "cited-by" are the only verbs you find acceptable, then define them as owl:inverseOf properties in an ontology. Honestly, I will help. Consider irw:isAbout and irw:isTopicOf as existing alternatives, though. Linked Data RWOs and Web Documents presumably serve the same purpose. Web standard solutions already exist for this part. > I understood the original post to be about the latter. E.g., if every > book, > chapter, section, and article actually had a DOI, then we could build a > doi[1] references doi[2] graph and be done with it. Since everything > doesn't > have a DOI, the question is in two parts: (a) how do we algorithmically > generate unique URIs in a way that guarantees preservation of the > identity > relationship, You should believe that URIs are names. It is impossible to express semantics adequately in a name. That's what HTTP GET and RDF are for. > and (b) how do we actually generate/store/query the > resulting > graph. Use Linked Data. Jeff > > Jodi, is any of this correct? > > > > On Tue, Jul 20, 2010 at 5:33 PM, Young,Jeff (OR) <[log in to unmask]> > wrote: > > > Stuart, > > > > Sorry, I didn't mean to discount citation representations along other > > content-negotiable dimensions. It seems likely that BCP-47 < > > http://tools.ietf.org/html/rfc5646> will eventually be upgraded to > > recognize signwriting. If so, my URI pattern suggestion could be > extended to > > support language, script, etc. like so: > > > > http://example.org/manifestation/1/citation-apa.{bcp-47}.txt > > > > In FRBR, serials are recognized as a distinct class so I assume this > URI > > pattern could be applied to suit all: > > > > http://example.org/serial/2/citation-apa.{bcp-47}.txt (text/plain) > > > > Jeff > > > > > -----Original Message----- > > > From: Code for Libraries [mailto:[log in to unmask]] On > Behalf Of > > > stuart yeates > > > Sent: Tuesday, July 20, 2010 4:14 PM > > > To: [log in to unmask] > > > Subject: Re: [CODE4LIB] "universal citation index" > > > > > > Young,Jeff (OR) wrote: > > > > http://example.org/manifestation/1/citation-apa.txt (text/plain) > > > > > > The problem I have with the use of (text/plain) is that too many > > > platforms still assume / default to latin1 for text/plain. While > this > > > appears to be reducing, with signwriting still coming through the > > > standards pipeline we're not out of the woods yet. > > > > > > And yes, there are serials in signwriting. > > > > > > cheers > > > stuart > > > -- > > > Stuart Yeates > > > http://www.nzetc.org/ New Zealand Electronic Text Centre > > > http://researcharchive.vuw.ac.nz/ Institutional Repository > > > > > > -- > Bill Dueber > Library Systems Programmer > University of Michigan Library