------------------------------------------------------------------------ Date: Tue, 23 Aug 2005 16:36:17 +0100 From: Andy Powell Subject: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ I had an action (possibly with Diane?) to put the proposed changes to DCMES definitions (the changes required in order to aling the definitions with the DCAM) into the form of a document (at least, I think that's what the action was!). I've done this and put the document in the DC RDF Taskforce's Wiki (since these changes are also very relevant to the work of that group). See: Proposed changes to DCMI property definitions http://www.ukoln.ac.uk/twiki/bin/view/Metadata/ProposedPropertyDefinitions Shout if you disagree with anything and/or have any other comments. Note that some of the proposed changes go a bit beyond a simple interpretation of what the DCAM says! ------------------------------------------------------------------------ Date: Tue, 23 Aug 2005 17:42:07 +0100 From: Pete Johnston Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ > Proposed changes to DCMI property definitions > http://www.ukoln.ac.uk/twiki/bin/view/Metadata/ProposedPropertyDefinitions > > Shout if you disagree with anything and/or have any other comments. The comment for dc:coverage includes === Coverage includes spatial location (a place name or geographic coordinates), temporal period (a period label, date, or date range) or jurisdiction (such as a named administrative entity). === I'm not sure "a place name" or (a set of) "geographic coordinates" _is_ a spatial location. I think they are means of identifying/describing/representing/labelling a spatial location. Same for "a period label". I think the form "named adminstrative entity" gets it right - the value is the entity, not its name. So maybe "a place name" -> "a named place" "geographic coordinates" -> "a location specified by (a set of?) geographic coordinates" "a period label" -> "a named period" Or rephrase the text in brackets to say "represented by [the names/labels etc]". Either way works as long as they are consistent, I think. The comment for dc:source includes === The present resource may be derived from the Source resource in whole or in part. Recommended best practice is to identify the referenced resource by means of a string or number conforming to a formal identification system. === This uses both "the Source resource" (yuk!) and "the referenced resource" to refer to the same thing, the value - maybe clearer to settle on "the referenced resource" (which is the form used in the descriptions of subprops of dc:relation, I think?) Also this and some of the other DCMES definitions talk about "the present resource" whereas the subprops of dc:relation talk about "the described resource" - the latter seems a bit more in keeping with the DCAM? The comment for dc:rights includes === If the Rights element is absent... === This phrasing does seem a bit at odds with the DCAM. I think it reflects a view in which "elements" are components in instances (as in XML (or LOM?)). But the DCAM has no view of "elements" being "present in" or "absent from" descriptions. A description contains statements _referring to_ the property or it doesn't. So I think this should really say something like (though neither of these seem quite right either!): > In the absence of a statement using the Rights property.... or maybe > In the absence of a Rights statement.... (though I recognise "Rights statement" might have other connotations) > I would recommend changing all usage of "[intellectual] content of the resource" to "resource" in the definitions I notice that you haven't proposed a change to the definition of dc:subject ("The topic of the content of the resource.") Not sure if that was deliberate? ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 14:23:41 +0100 From: Andy Powell Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ On Tue, 23 Aug 2005, Pete Johnston wrote: > So maybe > > "a place name" -> "a named place" > "geographic coordinates" -> "a location specified by (a set of?) > geographic coordinates" > "a period label" -> "a named period" Thanks. Fixed. > This uses both "the Source resource" (yuk!) and "the referenced > resource" to refer to the same thing, the value - maybe clearer to > settle on "the referenced resource" (which is the form used in the > descriptions of subprops of dc:relation, I think?) > > Also this and some of the other DCMES definitions talk about "the > present resource" whereas the subprops of dc:relation talk about "the > described resource" - the latter seems a bit more in keeping with the > DCAM? Thanks. Fixed. > So I think this should really say something like (though neither of > these seem quite right either!): > >> In the absence of a statement using the Rights property.... > > or maybe > >> In the absence of a Rights statement.... > > (though I recognise "Rights statement" might have other connotations) I've removed this sentence completely. >> I would recommend changing all usage of "[intellectual] content of the > resource" to "resource" in the definitions > > I notice that you haven't proposed a change to the definition of > dc:subject ("The topic of the content of the resource.") Not sure if > that was deliberate? I've only explicitly proposed removing "[intellectual] content of the" in those cases where I was making changes to the term for other reasons. I missed subject off the list of possible problems by accident. I've now added it. Thanks. ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 14:33:36 +0100 From: Andy Powell Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ I have also added a proposed change for Language (i.e. removal of the 'eng' example from the comment) as noted in previous minutes. http://www.ukoln.ac.uk/twiki/bin/view/Metadata/ProposedPropertyDefinitions ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 15:29:30 +0100 From: Andy Powell Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ On Wed, 24 Aug 2005, Andy Powell wrote: > I have also added a proposed change for Language (i.e. removal of the 'eng' > example from the comment) as noted in previous minutes. > > http://www.ukoln.ac.uk/twiki/bin/view/Metadata/ProposedPropertyDefinitions I also note from the Washington minutes that the label for Subject was supposed to change - so I have also added a proposed change for Subject. ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 10:33:57 -0400 From: "Diane I. Hillmann" Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Andy, It was probably me, but whoosh, the summer has flown by. In any case, thanks for following through on these--and in general I agree with your suggestions (with Pete's additions). I have a comment about Coverage, however. I'm a bit wary of using "topic" here, since it creates potential confusion with Subject. I would prefer "scope" (and I'm not sure why you think that creates a conflict with format and extent?) since there's not necessarily an "aboutness" implied for coverage that the word "topic" suggests. Particularly with the comment about jurisdictional coverage, "topic" is problematic. I'd be willing to compromise with "scope or topic" if necessary ... ;-) Particularly because many of our users come to DC from MARC, this is an issue that comes up over and over. MARC handles the differences between "topic" and "scope" by placement of geographic information within a subject string. Resources "about" a place or time have the geographic or temporal piece first in the subject string; where the scope of the topic is limited by geographic or temporal aspects, it comes after the topic. For historical reasons, we've treated Coverage differently, but, even so, it behooves us NOT to blur the difference by emphasizing only the topical uses of Coverage. Diane ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 15:48:31 +0100 From: Andy Powell Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ On Wed, 24 Aug 2005, Diane I. Hillmann wrote: > It was probably me, but whoosh, the summer has flown by. In any case, thanks > for following through on these--and in general I agree with your suggestions > (with Pete's additions). I have a comment about Coverage, however. > > I'm a bit wary of using "topic" here, since it creates potential confusion > with Subject. I would prefer "scope" (and I'm not sure why you think that > creates a conflict with format and extent?) since there's not necessarily an > "aboutness" implied for coverage that the word "topic" suggests. Particularly > with the comment about jurisdictional coverage, "topic" is problematic. I'd > be willing to compromise with "scope or topic" if necessary ... ;-) > > Particularly because many of our users come to DC from MARC, this is an issue > that comes up over and over. MARC handles the differences between "topic" and > "scope" by placement of geographic information within a subject string. > Resources "about" a place or time have the geographic or temporal piece first > in the subject string; where the scope of the topic is limited by geographic > or temporal aspects, it comes after the topic. For historical reasons, we've > treated Coverage differently, but, even so, it behooves us NOT to blur the > difference by emphasizing only the topical uses of Coverage. Hmmm... interesting. I'm not sure I understand what you mean by 'scope' here. I thought that Coverage was to do with the 'aboutness' of the resource. In fact, I've always worked on the basis that Coverage is, essentially, a sub-property of Subject. (Though of course, as good DCers we could never say such a thing out loud! :-) ). This is clearly not your view... So, some examples. I've always assumed that when we say dc:coverage: bath, uk we mean that the topic of the resource is 'bath, uk' - the resource is about 'bath, uk'. And further, that we could just as well say dc:subject: bath, uk but we are being a little more specific by using Coverage. Similarly, when we say dc:coverage: 19th century we mean that the resource is about the 19th century - i.e. the topic of the resource is "19th century". What do you mean by those two statements (assuming that it isn't the same as my intepretation)? I presume that you mean something different by 'scope' (as opposed to 'topic') - something like "is only relevent in this context" perhaps? So that, for example, if for some bizarre reason we wanted to say that a resource is "about the US, but only appropriate for use by people in the UK' we'd say dc:subject: US dc:coverage: UK ?? That certainly isn't my understanding of what Coverage is about. ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 11:02:28 -0400 From: "Diane I. Hillmann" Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Andy: One example of what I'm talking about is where, for instance, a government site might have resources that apply to only a particular jurisdiction or sub-jurisdiction. The resources are not really "about" that jurisdiction or place, but the jurisdiction within which the resource is applicable would be indicated in Coverage. There might also be a Subject, which might categorize the general topic of the resource. The purposes of these two elements are quite different, and not both relating to "aboutness" as I understand it. On the other hand, if you have resources for tourists about that jurisdiction or place, they are certainly about that place. Diane ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 16:18:58 +0100 From: Andy Powell Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ On Wed, 24 Aug 2005, Diane I. Hillmann wrote: > Andy: > > One example of what I'm talking about is where, for instance, a government > site might have resources that apply to only a particular jurisdiction or > sub-jurisdiction. The resources are not really "about" that jurisdiction or > place, but the jurisdiction within which the resource is applicable would be > indicated in Coverage. There might also be a Subject, which might categorize > the general topic of the resource. The purposes of these two elements are > quite different, and not both relating to "aboutness" as I understand it. OK, thanks. Blimey, I think my understanding of Coverage has been wrong up until now! :-( Just to confirm... if I say dc:coverage: UK I am *not* saying that the resource is about the UK. Rather, I am saying that the resource is only applicable in the UK (i.e. only appropriate for use in the UK)? Or am I saying something else. > On the other hand, if you have resources for tourists about that jurisdiction > or place, they are certainly about that place. In which case, I presumably use Subject: dc:subject: UK Is this what you are saying? Is this how everyone uses Coverage? ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 08:30:36 -0700 From: Stuart Sutton Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ My understanding of coverage is exactly as Andy states it. ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 08:36:50 -0700 From: Stuart Sutton Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Hmmm. Diane, this runs absolutely counter to my assumptions about the semantics of coverage (per my response to Andy's earlier response). In fact, I am working on a project that has defined a local property for jurisdiction (area/zone of mandated applicability) for academic standards in teaching and learning in the US because the focus on dc:coverage on "aboutness" made it an inappropriate place to express such notions of jurisdiction. ------------------------------------------------------------------------ Date: Wed, 24 Aug 2005 11:56:57 -0400 From: "Diane I. Hillmann" Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Well, I think this discussion points up the possible downside of making explicit the topicality of Coverage, which I really don't think was intended. But maybe I'm the minority in this. I certainly think that it has been used by may to express "aboutness" but whether that means we should limit it to that and force implementors to create local properties to express applicability is something we might want to discuss further. It seems to me that, if we want to allow both uses, we might consider a broader definition, and suggest to somebody that they propose a refinement for coverage? (I'm ducking here to avoid the rotten tomatoes) ------------------------------------------------------------------------ Date: Tue, 30 Aug 2005 09:41:31 +0100 From: Andrew Wilson Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Sorry for the lateness of this reply - I had to make an unexpected trip to Australia for a family medical emergency. I do want to make a comment about Coverage. My understanding of its use corresponds to Andy's (and Stuart's). Would it make sense to change the word 'under' in the proposed definition to 'to', so it reads "...or the jurisdiction to whcih the resource is relevant"? ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 13:33:31 +0100 From: Pete Johnston Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Andrew said: > I do want to make a comment about Coverage. My understanding > of its use corresponds to Andy's (and Stuart's). Would it > make sense to change the word 'under' in the proposed > definition to 'to', so it reads "...or the jurisdiction to > whcih the resource is relevant"? I'm a bit confused now. ;-) I think Stuart and Andy(?) are suggesting that the value space of dc:coverage excludes "jurisdiction to which the resource is relevant" and includes _only_ (spatial/temporal) "aboutness". So _if_ that is the case - and I don't know whether it is the intention behind dc:coverage or not: I've always been a bit confused about the way people seem to use dc:coverage! - then it seems to me the definition should _exclude_ that reference. And I'd argue that we should also say that dc:coverage is a subproperty of dc:subject, but I'm aware I won't get very far making that argument... ;-) However, I do note that AGLS seems to take the view that the value space of dc:coverage _includes_ spatial/temporal relevance/applicability (or however we describe it!) See AGLS http://www.naa.gov.au/recordkeeping/gov_online/agls/metadata_element_set .html where agls:jurisdiction is a refinement of dc:coverage with: Definition: The name of the political/administrative entity covered by the content of the resource. Comment: Jurisdiction is a description of the territory over which a particular government exercises its authority or a particular business transacts its operations, to which the resource content is applicable. If the value space of dc:coverage includes both spatial/temporal "aboutness" _and_ spatial/temporal relevance/applicability - which I think is Diane's position (?) and seems to be reflected in AGLS usage (and other e-gov usage? I'm not sure) at least - then the definition should make that clear. And in that case dc:coverage is not a subproperty of dc:subject - though I could have a spatialTopic property which was a subproperty of both dc:subject and dc:coverage! ;-) ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 15:26:34 +0100 From: Andy Powell Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ On Wed, 31 Aug 2005, Pete Johnston wrote: > Andrew said: > >> I do want to make a comment about Coverage. My understanding >> of its use corresponds to Andy's (and Stuart's). Would it >> make sense to change the word 'under' in the proposed >> definition to 'to', so it reads "...or the jurisdiction to >> whcih the resource is relevant"? > > I'm a bit confused now. ;-) > > I think Stuart and Andy(?) are suggesting that the value space of > dc:coverage excludes "jurisdiction to which the resource is relevant" > and includes _only_ (spatial/temporal) "aboutness". That would be nice... but that wasn't really what I was suggesting. My proposed (new) definition is The spatial or temporal scope/topic(?) of the resource or the jurisdiction under which the resource is relevant. i.e. I don't exclude "jurisdiction to which the resource is relevant". I don't see how we can remove the bit about jurisdiction without breaking a lot of existing usage. Basically, it seems to me, that coverage is pretty much as broken as format is (since they both try to define two things :-( ). But we are stuck with that situation. I'm arguing that (however much I dislike it!) coverage covers both 'aboutness' (the spatial or temporal scope/topic(?) of the resource) and 'applicability' (the jurisdiction under which the resource is relevant). Furthermore, I thought that Diane was arguing that coverage only covers the 'applicability' aspect, i.e. the applicability in spatial, temporal or jusidictional terms, and *not* the 'aboutness'... but maybe I'm mis-interpretting what she is saying. > So _if_ that is the case - and I don't know whether it is the intention > behind dc:coverage or not: I've always been a bit confused about the way > people seem to use dc:coverage! - then it seems to me the definition > should _exclude_ that reference. And I'd argue that we should also say > that dc:coverage is a subproperty of dc:subject, but I'm aware I won't > get very far making that argument... ;-) > > However, I do note that AGLS seems to take the view that the value space > of dc:coverage _includes_ spatial/temporal relevance/applicability (or > however we describe it!) > > See AGLS > http://www.naa.gov.au/recordkeeping/gov_online/agls/metadata_element_set > .html > > where agls:jurisdiction is a refinement of dc:coverage with: > > Definition: The name of the political/administrative entity covered by > the content of the resource. > Comment: Jurisdiction is a description of the territory over which a > particular government exercises its authority or a particular business > transacts its operations, to which the resource content is applicable. > > If the value space of dc:coverage includes both spatial/temporal > "aboutness" _and_ spatial/temporal relevance/applicability - which I > think is Diane's position (?) and seems to be reflected in AGLS usage > (and other e-gov usage? I'm not sure) at least - then the definition > should make that clear. Well, as I say above, that *is* also my position (though I didn't think it was Diane's) and the proposed definition makes both aspects of the semantics clear (to me). > And in that case dc:coverage is not a > subproperty of dc:subject OK, well I'm happy to go with that interpretation. However, FWIW, my view is that "the jurisdiction under which the resource is relevant" is close enough to 'aboutness' to fall within dc:subject. For example, I'm struggling to think of an example where the jurisdiction (as given in coverage) wouldn't also be appropriate for adding to a 'Keywords' meta tag. > - though I could have a spatialTopic property > which was a subproperty of both dc:subject and dc:coverage! ;-) Yes. What a mess :-( ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 16:04:21 +0100 From: Pete Johnston Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Ansy said: > That would be nice... but that wasn't really what I was > suggesting. My > proposed (new) definition is > > The spatial or temporal scope/topic(?) of the resource or the > jurisdiction under which the resource is relevant. > > i.e. I don't exclude "jurisdiction to which the resource is relevant". Ah, OK. > I don't see how we can remove the bit about jurisdiction > without breaking a lot of existing usage. Yes, I agree. That was my concern too. > Basically, it seems to me, that coverage is pretty much as broken as > format is (since they both try to define two things :-( ). > But we are stuck with that situation. Yes. Sigh. :-( * wishes we could just start again * [snip] > OK, well I'm happy to go with that interpretation. However, > FWIW, my view > is that "the jurisdiction under which the resource is > relevant" is close enough to 'aboutness' to fall within dc:subject. That may apply in some cases but I'm not sure it works in all cases. If I have (a) "Planning your tour of the UK (for visitors from the US)" (b) "Planning your tour of the UK (for vistors from Brazil)" the spatial relevance/applicability of (a) is USA (or maybe USA and UK) and the spatial relevance/applicability of (b) is Brazil (or maybe Brazil and UK) But the spatial "aboutness" of both (a) and (b) is UK, not USA or Brazil. > For example, I'm > struggling to think of an example where the jurisdiction (as given in > coverage) wouldn't also be appropriate for adding to a > 'Keywords' meta tag. Ah, but I think meta/keywords is broader than dc:subject? For my (a) above, I might well include both USA and UK as my meta/keywords but that doesn't necessarily mean I should include them both as values for dc:subject. Actually now that I look for it I'm not sure where the scope of meta/keywords is specified! The best I can find is http://www.w3.org/TR/html401/struct/global.html#edef-META and http://www.w3.org/TR/html401/appendix/notes.html#recs It seems to me the keywords in meta/keywords aren't limited to aboutness: they are _any_ keywords useful for retrieval. Whereas for dc:subject, keywords are supposed to describe the topic of the resource i.e, they are limited to "aboutness", not any old keyword. But yeah, I am hair-splitting here, and in practice I suspect the use of dc:subject is rather looser too. ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 10:25:42 -0700 From: Stuart Sutton Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Well, if coverange is capable of being both aboutness and jurisdiction, it is quite broken indeed. People looking for resources about "California" using coverage are unlikely to be satisfied retrieving descriptions of the various chapters of the California statutes, administrative code, and case law. And clearly, people looking for jurisdictional documents would be absolutely unlikely to be satisfied in the least with resources on the history of the California state flower (poppy) or the Gold Rush. Jurisdiction and aboutness are conceptually different notions. To be capable of expressing both jurisdiction and aboutness is to end up expressing neither with any certainty...or so it seems to me. ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 13:39:02 -0400 From: "Diane I. Hillmann" Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Stuart, et al.: I do believe that this ambiguity (or brokenness, if you insist) has been with us since he beginning, though perhaps it took the messiness of implementation to bring it to the fore. I do believe that, having been around at the inception (such as it was), this was not unintentional, though indeed our current sensibility is very much against the kind of ambiguity represented here. The inclusion of the Coverage element was pushed by the folks for whom geographic representation was important--I don't think they cared so much about the definition of the element, just that there was someplace to put geographic information, for whatever purpose. I agree with Stuart that the use cases we have discussed, and the implementations we know about, suggest that we might want to take some action here. I agree with Pete (I think it was) that there is enough "aboutness" in our notion of jurisdiction that we should be able to do something reasonable, without breaking existing implementations. It may be that we should look to Format as our guide, where the Extent and Medium refinements make the distinctions we discovered "after the fact" of defining the Format element. In the case of Coverage, we might be able to do with one. And yes, this was what I was trying to get at in my first post, though certainly, in retrospect, I didn't get there via a straight line. But then, neither did much in Dublin Core ... ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 19:47:17 +0100 From: Pete Johnston Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ > I do believe that this ambiguity (or brokenness, if you insist) has > been with us since he beginning, though perhaps it took the messiness > of implementation to bring it to the fore. I do believe that, having > been around at the inception (such as it was), this was not > unintentional, though indeed our current sensibility is very much > against the kind of ambiguity represented here. The inclusion of the > Coverage element was pushed by the folks for whom geographic > representation was important--I don't think they cared so much about > the definition of the element, just that there was someplace to put > geographic information, for whatever purpose. > > I agree with Stuart that the use cases we have discussed, and the > implementations we know about, suggest that we might want to take > some action here. I agree with Pete (I think it was) that there is > enough "aboutness" in our notion of jurisdiction that we should be > able to do something reasonable, without breaking existing > implementations. I think that was Andy. I was agreeing with Stuart and saying that "aboutness" and "relevance/applicability" were different notions (and by implication agreeing that having them rolled into a single property didn't seem very useful). What I wasn't sure about was whether the dc:coverage property did actually encompass both these notions, but it seems that it does. :-( > It may be that we should look to Format as our > guide, where the Extent and Medium refinements make the distinctions > we discovered "after the fact" of defining the Format element. In the > case of Coverage, we might be able to do with one. I guess it would be possible to coin something like dcterms:relevance ("A time or place for which the resource is relevant or applicable") with dcterms:relevance rdfs:subpropertyOf dc:coverage (I think you could get away without dcterms:spatialRelevance and dcterms:temporalRelevance as well, and examine the type of the value instead, but I'm not sure.) But that doesn't change the fact that you can't tell whether references to the existing dc:coverage, dcterms:spatial and dcterms:temporal properties are being used for aboutness or for relevance. i.e. if I find thing:x dcterms:spatial country:Spain . thing:x dcterms:relevance country:Spain . I still don't know whether thing:x is "about" Spain or not. ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 16:47:57 -0400 From: "Diane I. Hillmann" Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ >I think that was Andy. I was agreeing with Stuart and saying that >"aboutness" and "relevance/applicability" were different notions (and by >implication agreeing that having them rolled into a single property >didn't seem very useful). What I wasn't sure about was whether the >dc:coverage property did actually encompass both these notions, but it >seems that it does. :-( Oops, sorry! Sometimes when you guys get the bit between your teeth, it's hard to keep up ... ;-) > > It may be that we should look to Format as our >> guide, where the Extent and Medium refinements make the distinctions >> we discovered "after the fact" of defining the Format element. In the >> case of Coverage, we might be able to do with one. > >I guess it would be possible to coin something like > >dcterms:relevance ("A time or place for which the resource is relevant >or applicable") > >with > >dcterms:relevance rdfs:subpropertyOf dc:coverage Given that we've not seen much call for relevance for temporal information, and "jurisdiction" seems to be our challenge, couldn't we consider some of the definitional work that Stuart has done for that as an element and use is for a refinement? Really, doesn't Spatial and Temporal deal primarily with the "aboutness" portion? >(I think you could get away without dcterms:spatialRelevance and >dcterms:temporalRelevance as well, and examine the type of the value >instead, but I'm not sure.) > >But that doesn't change the fact that you can't tell whether references >to the existing dc:coverage, dcterms:spatial and dcterms:temporal >properties are being used for aboutness or for relevance. > >i.e. if I find > >thing:x dcterms:spatial country:Spain . >thing:x dcterms:relevance country:Spain . > >I still don't know whether thing:x is "about" Spain or not. > That's true, but isn't that a function of the fact that "relevance" as a concept isn't particularly well understood in our community? In contrast, I think the people who work with jurisdictions (like lawyers and governments) understand that concept instinctively and don't tend to ask questions like the one you just asked ... ;-) As for the worry about already encoded information, consider when we added StillImage and MovingImage to DCMIType. We didn't know whether things that were already encoded as images were one or the other. Diane (keeping her head down, sort of) ------------------------------------------------------------------------ Date: Thu, 1 Sep 2005 08:45:56 +0100 From: Pete Johnston Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Quoting "Diane I. Hillmann" : > Given that we've not seen much call for relevance for temporal > information, and "jurisdiction" seems to be our challenge, couldn't > we consider some of the definitional work that Stuart has done for > that as an element and use is for a refinement? Really, doesn't > Spatial and Temporal deal primarily with the "aboutness" portion? I was assuming that both dcterms:spatial and dcterms:temporal suffered from the same problem as dc:coverage (i.e. they encompass both "aboutness" and "relevance/applicability"). I'm afraid "primarily" isn't any use to an application, though ;-). Either they do or they don't. > That's true, but isn't that a function of the fact that "relevance" > as a concept isn't particularly well understood in our community? In > contrast, I think the people who work with jurisdictions (like > lawyers and governments) understand that concept instinctively and > don't tend to ask questions like the one you just asked ... ;-) I hadn't really developed the concept of "relevance" in any real way - what I was trying to get at was that since dc:coverage has both a temporal and spatial aspect, the problem of distinguishing aboutness and relevance is present in both of those dimensions. (Of course then there is an overlap with things like dcterms:valid, too) (To be honest, it's a bit of a mess: I suspect the problem isn't really fix-able without scrapping the current dc:coverage term and redefining it more coherently, but I recognise that would be a "major semantic impact"/new URI needed sort of case, so I'm not really proposing that as a course of action.) > As > for the worry about already encoded information, consider when we > added StillImage and MovingImage to DCMIType. We didn't know whether > things that were already encoded as images were one or the other. Agreed, but in that case the more general case ("it's an Image") is still (fairly) useful; in the dc:coverage case (as Stuart pointed out yesterday), the statement that "It's either about this time/place or it's relevant/applicable to this time/place - but you can not know which" is, arguably, not very useful at all. You might almost just as well use dc:relation (some association with thsi time/place). This has always been my problem with dc:coverage - the _potential_ scope means that any _actual_ use serves little real purpose. ------------------------------------------------------------------------ Date: Wed, 31 Aug 2005 19:03:39 +0100 From: Pete Johnston Subject: Re: Proposed changes to DCMI property definitions To: DC-USAGE@JISCMAIL.AC.UK ------------------------------------------------------------------------ Stuart said: > Jurisdiction and aboutness are conceptually > different notions. To be capable of expressing both > jurisdiction and aboutness is to end up expressing neither > with any certainty...or so it seems to me. I'm inclined to agree with Stuart, and in an ideal world the Right Thing To Do would be to coin different properties to represent the different notions. But given that the existing hybrid use of all _three_ of dc:coverage, dcterms:spatial and dcterms:temporal is out there, the options for fixing it seem pretty messy.