Cross-posting to oslc-cm a discussion on the Helios_bt (bug ontology). Discussion of usage of dcterms:title and dcterms:abstract. I have enclosed my latest response as <Steve>
Thanks, Steve Speicher | IBM Rational Software | (919) 254-0645 > ---------- Forwarded message ---------- > From: Steve Speicher > Date: Mon, Jul 26, 2010 at 7:17 AM > Subject: Re: [helios:wp3] Some initial comments on Helios_bt, > aligning usage of DC and OSLC-CM > To: Olivier Berger > > > On Mon, Jul 26, 2010 at 5:08 AM, Olivier Berger <[email protected] > > wrote: > Le jeudi 22 juillet 2010 à 09:52 -0400, Steve Speicher a écrit : > > These comments are on > > http://heliosplatform.sourceforge.net/ontologies/2010/05/helios_bt.html > > > > Within OSLC-CM [1], we've re-evaluated our usage of some Dublin Core > > properties [2]. Namely, we felt we were using dc:title when > > dc:abstract may be more appropriate. We have looked at utilizing > > dc:title for a shorter form for a change request. > Thanks for your feedback. > > I'm a bit dubious on the (title) vs (title + abstract). > > Refering to academic publishing for instance, I see title of a paper as > a descriptive phrase (short) of the content, whereas abstract would be a > couple paragraphs explaining in more details the meaning of the title. > It should convey more than an identifier (for which there's > dc:identifier ;) > > Now, in the example you provide below, I think that "Issue58365" is not > meaningful at all. It's more an identifier than the title of a bug > report, for me. Even though the page at > <http://qa.mandriva.com/show_bug.cgi?id=58365> may actually be titled so > (actually, here the html:title is in indeed "Bug 58365 - mantis uses > bundled libraries instead of system-wide ones", as some RDFa tools > extract it...) > > Maybe the abstract could be some kind of a summary of the various > aspects of the bug report, as added once the support staff has > summarized an abstract, after several iterations with reporter and other > concerned parties ? > In the Debian bugtracker (debbugs) there's the "summary" field [0] that > may be set for just such use. > > In absence of the need for such an abstract/summary, should the title be > descriptive enough, the title only could be used. > > So I would rather see something like : > > # This is turtle format for RDF > @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>. > > @prefix dcterms: <http://purl.org/dc/terms/>. > @prefix oslc_cm: <http://open-services.net/ns/cm#>. > > <http://qa.mandriva.com/show_bug.cgi?id=58365> > > a oslc_cm:ChangeRequest; > dcterms:identifier "58365"; > dcterms:title "mantis uses bundled libraries instead > of system-wide ones"; > dcterms:abstract "mantis uses libraries like nusoap, > adodb and phpmailer, which should not be shipped along the ."; > > [0] http://www.debian.org/Bugs/server-control#summary > > <Steve> When looking at this, I tried using the academic paper and > publication analogy (while keeping to what DC describes for the terms). > > When I look at title, it is something that doesn't identify a > publication but a meaningful title say "REST best practices". As > you state an abstract is at most a couple paragraphs. > Though I believe in most bug tools I've seen, there is no concept of > a "couple paragraph" abstract more true to what DC calls the > abstract which is a "summary of the resource". I see a bug title > being, like academic papers, a meaningful way to refer to the > resource (perhaps not a UUID or URL) but something that would make > sense to humans. For example, when I tell a colleague what bug to > refer to I say "Bug123". > The dc:identifier is like the ISBN for a publication. It is not > necessarily a URI, though the service knows that it is a way to > identify the resource. </Steve> > Maybe this should be discussed on OSLC side (feel free to quote me /forward). > > Yes, I think I'll forward to OSLC mailing list. I will also forward > to a couple of people to get an opinion on. > > > (btw, in OSLC we use DCTERMS and not DCELEMENTS) > > > Uh... can you update me : just the latest version of DC properties that > are now called terms and using http://purl.org/dc/terms/ instead of > previous one that used http://purl.org/dc/elements/ but same semantics ? > > <Steve> Yes, DCTERMS has different namespace and includes all the DCELEMENTS > with the same semantics. In OSLC we depend on additional terms such > as dcterms:modified. </Steve> >
