Print

Print


Thanks for this answer Kevin

The examples you give of using MADS here are, as you note, not too far away from what I was asking for - and so look good to me. I think some documentation and 'best practice' suggesting this, or giving the outline as suggested here, would be a really useful resource for those working with LCSH in a linked data context.

Thanks again for this full and useful answer

Best wishes,

Owen

Owen Stephens
Owen Stephens Consulting
Web: http://www.ostephens.com
Email: [log in to unmask]
Telephone: 0121 288 6936

On 9 Nov 2011, at 22:55, Ford, Kevin wrote:

> Dear Owen,
> 
> Thanks for writing.
> 
>> As it has been added to the catalogue, I assume it is a valid heading,
>> so we check for a URI for this heading from id.loc.gov - but there
>> isn't one, because although it may be valid, it isn't authorised, and
>> so doesn't fall into the scope of the data published on id.loc.gov
> -- And there's the rub.  ID provides identifiers for LCSH headings (single or pre-coordinated) for which there is a MARC record.  It is basically possible to create an infinite number of perfectly-valid, pre-coordinated LC subject headings.  I've started down this road- there are roughly 6.2 unique subject headings in the LC bibliographic dataset - but I don't know how it ends.  It is a good thing to highlight, however, because it is a something that many do not necessarily consider when starting out with ID.LOC.GOV.
> 
>> There are some parts of what we did that I suspect are open to
>> discussion as to their validity - for example, is it valid to apply the
>> URI http://id.loc.gov/authorities/sh85118587#concept (=Science--Study
>> and Teaching) when the original heading was Science--Study and
>> Teaching--Research.
> -- Yea, we had this debate too.  Decided, in the end, to do identify each component separately.
> 
>> I can see that MADS might offer this type of approach - but this seems
>> overly complex for what I'm looking to achieve - I don't need to
>> express an authority file, but rather relationships to existing
>> resources.
> -- Yes and no.  Certainly no more difficult if you are already considering creating a local URI for a pre-coordinated LCSH that you can't match at ID.  And, if you are already using ID URIs, there's been at least a tacit acceptance of the additonal data (whether in SKOS or MADS/RDF) since most of the resources at ID contain substantial information.  In reality, however, very little is needed to express a pre-coordinated heading in MADS/RDF: a label and componentList are sufficient to capture your intent and meet your need.  Using you local URI example:
> 
> @prefix madsrdf: <http://www.loc.gov/mads/rdf/v1#>.
> @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.
> 
> <http://data.open.ac.uk/topic/library/science--study_and_teaching--research> madsrdf:authoritativeLabel "Science--Study and teaching--Research";
> 	madsrdf:componentList (<http://id.loc.gov/authorities/subjects/sh85118553> <http://id.loc.gov/authorities/subjects/sh2001008697> <http://id.loc.gov/authorities/subjects/sh2002006576>);
> 	a madsrdf:ComplexSubject.
> 
> Currently the madsrdf:componentList property has a range of rdf:List.  In a future update I anticipate the range of madsrdf:componentList will be an rdf:List OR rdf:Seq.  This will simplify querying, but it does not significantly alter the construction (I've provided an RDF/XML serialization of the above example and the following example at the end of this email):
> 
> <http://data.open.ac.uk/topic/library/science--study_and_teaching--research> madsrdf:authoritativeLabel "Science--Study and teaching--Research";
> 	madsrdf:componentList _:bnode1302373376;
> 	a madsrdf:ComplexSubject.
> 
> _:bnode1302373376 rdf:_1 <http://id.loc.gov/authorities/subjects/sh85118553>;
> 	rdf:_2 <http://id.loc.gov/authorities/subjects/sh2001008697>;
> 	rdf:_3 <http://id.loc.gov/authorities/subjects/sh2002006576>;
> 	a rdf:Seq.
> 
> For your troubles, you get the pre-formatted label *and* the component resources of the label.  Naturally, it is possible to construct a label without representing it natively in the data, but it may be of benefit with regard to indexing and supplying the label reduces the need to perform look-ups.  After all, when you query the resource of a single term heading, you're looking for its label.  It seems logical to extend the same principle to pre-coordinated ones.
> 
>> Hopefully I'm not saying anything very stupid ... thoughts/comments
>> welcome
> -- Absolutely and categorically, no.  It's really very helpful for us to hear about *how* the data is being used and what challenges users face using the data.  I've said it before on this list and I'll say it again: hearing these accounts, receiving questions, and getting feedback inform our priorities and service enhancements.  Your account is very much welcome.  Thanks again.
> 
> Warmly,
> 
> Kevin
> 
> 
> *Using an rdf:List*
> <rdf:RDF 
>  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>  xmlns:madsrdf="http://www.loc.gov/mads/rdf/v1#">
> 
> 	<madsrdf:ComplexSubject rdf:about="http://data.open.ac.uk/topic/library/science--study_and_teaching--research">
> 		<madsrdf:authoritativeLabel>Science--Study and teaching--Research</madsrdf:authoritativeLabel>
> 		<madsrdf:componentList rdf:parseType="Collection">
> 			<madsrdf:Topic rdf:about="http://id.loc.gov/authorities/subjects/sh85118553"/>
> 			<madsrdf:Topic rdf:about="http://id.loc.gov/authorities/subjects/sh2001008697"/>
> 			<madsrdf:Topic rdf:about="http://id.loc.gov/authorities/subjects/sh2002006576"/>
> 		</madsrdf:componentList>
> 	</madsrdf:ComplexSubject>
> 
> </rdf:RDF>
> 
> *Using an rdf:Seq*
> <rdf:RDF 
>  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>  xmlns:madsrdf="http://www.loc.gov/mads/rdf/v1#">
> 
> 	<madsrdf:ComplexSubject rdf:about="http://data.open.ac.uk/topic/library/science--study_and_teaching--research">
> 		<madsrdf:authoritativeLabel>Science--Study and teaching--Research</madsrdf:authoritativeLabel>
> 		<madsrdf:componentList>
> 			<rdf:Seq>
> 				<rdf:li rdf:resource="http://id.loc.gov/authorities/subjects/sh85118553"/>
> 				<rdf:li rdf:resource="http://id.loc.gov/authorities/subjects/sh2001008697"/>
> 				<rdf:li rdf:resource="http://id.loc.gov/authorities/subjects/sh2002006576"/>
> 			</rdf:Seq>
> 		</madsrdf:componentList>
> 	</madsrdf:ComplexSubject>
> 
> </rdf:RDF>
> 
> 
> 
>> -----Original Message-----
>> From: Ford, Kevin
>> Sent: Wednesday, November 09, 2011 4:50 PM
>> To: Ford, Kevin
>> Subject: RE: [ID.LOC.GOV] Linking LCSH
>> 
>> 
>> 
>> From: Authorities and Vocabularies Service Discussion List
>> [mailto:[log in to unmask]] On Behalf Of Owen Stephens
>> Sent: Tuesday, November 08, 2011 2:37 PM
>> To: [log in to unmask]
>> Subject: [ID.LOC.GOV] Linking LCSH
>> 
>> Earlier this year I worked on a project[1] to create Linked Data
>> representation of a small set of library resources at The Open
>> University in the UK. I've been meaning to write up some thoughts on
>> the issues we encountered in linking to the LCSH representations on
>> id.loc.gov, and have finally got round to it, and would be interested
>> in feedback from others on our experience and my thoughts/reflections -
>> and of course if I'm making any obvious mistakes in my interpretation.
>> 
>> I presented some of these thoughts at a Linked Data and Libraries event
>> earlier this year - my slides and notes are
>> at http://www.slideshare.net/ostephens/linking-lcsh-and-other-
>> stuff from page 7 onwards.
>> 
>> We were working from MARC21 records, and focussing on the 'topical'
>> subject headings. So, typically our starting data might look something
>> like:
>> 650 #0 $aScience$xStudy and Teaching$xResearch OR Science--Study and
>> Teaching--Research
>> 
>> As it has been added to the catalogue, I assume it is a valid heading,
>> so we check for a URI for this heading from id.loc.gov - but there
>> isn't one, because although it may be valid, it isn't authorised, and
>> so doesn't fall into the scope of the data published on id.loc.gov
>> 
>> We could at this point just stop, and not provide an external link -
>> but this seemed like a wasted opportunity - we wanted to provide
>> external links where possible, and in this case there are actually URIs
>> on id.loc.gov for parts of the subject heading, just not the whole
>> thing. So checking we can find URIs for the following:
>> 
>> Science--Study and Teaching
>> = http://id.loc.gov/authorities/sh85118587#concept
>> Science (as Topical Term)
>> = http://id.loc.gov/authorities/sh85118553#concept ** Study and
>> Teaching (as a general subdivision)
>> = http://id.loc.gov/authorities/sh2001008697#concept
>> Research (as a general subdivision)
>> = http://id.loc.gov/authorities/sh2002006576#concept **
>> 
>> We have all the components we need to build the equivalent of the
>> original 650 field but using URIs instead of textual strings - and it
>> seems a shame not to use these.
>> 
>> In our project we decided to just allocate as many relevant URIs as we
>> could - and in order to preserve the full, but not authorised, LCSH we
>> also coined a local URI for this - you can see the results
>> at http://data.open.ac.uk/page/library/289148
>> 
>> There are some parts of what we did that I suspect are open to
>> discussion as to their validity - for example, is it valid to apply the
>> URI http://id.loc.gov/authorities/sh85118587#concept (=Science--Study
>> and Teaching) when the original heading was Science--Study and
>> Teaching--Research. However, what I think is missing is a better way of
>> using the LCSH component parts to express the LCSH as linked data. My
>> first thoughts are that it might be possible to use a mechanism like
>> the Bibliographic Ontology BibO uses for lists of authors, which is to
>> have an 'authorList' as well as individual authors[2], which allows you
>> to specify and ordered list alongside its component parts.
>> 
>> In this case you could do something like:
>> lcsh:headingList
>> ( <http://id.loc.gov/authorities/sh85118553#concept> <http://id.loc.gov
>> /authorities/sh2001008697#concept> <http://id.loc.gov/authorities/sh200
>> 2006576#concept>)
>> 
>> This would allow full use of the URIs already published on id.loc.gov,
>> and offer more links from library data.
>> 
>> I can see that MADS might offer this type of approach - but this seems
>> overly complex for what I'm looking to achieve - I don't need to
>> express an authority file, but rather relationships to existing
>> resources.
>> 
>> Hopefully I'm not saying anything very stupid ... thoughts/comments
>> welcome
>> 
>> 1. http://lucero-project.info/lb
>> 2. http://bibliontology.com/content/article
>> ** An aside - it isn't (or at least, wasn't) possible to search for
>> 'topical terms' separate to 'general subdivisions' so for headings like
>> 'Science' or 'Research' this creates a problem - it would be really
>> useful to be able to search for topical terms separate to general
>> subdivisions in order to identfy the correct URI
>> 
>> Owen Stephens
>> Owen Stephens Consulting
>> Web: http://www.ostephens.com
>> Email: [log in to unmask]
>> Telephone: 0121 288 6936