Mac,
This work has been my vocation, avocation, and now… seeping into my vacation. <Steven> a <DullBoy> . This is my last email while on vacation. :)
Comments below.
Thanks,
Steven
————
Steven Folsom
Metadata Strategist and Standards Advocate
Cornell University Library
On 8/8/15, 6:40 PM, "Thomas Berger" <[log in to unmask]> wrote:
>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>Am 08.08.2015 um 19:33 schrieb Karen Coyle:
>> On 8/7/15 7:38 PM, Steven Folsom wrote:
>>> Sorry, just realized Series/Part relationship should be at the BF:Wor
>k
>>> level. The Work is the Part. Instances manifestations of those parts.
>>
>> but transcribed information is about the Instance, not the work. So
>> you've identified another distinction between transcription of series
>> statements and linking. As long as we put information about the series
>> in the Instance graph, we are not showing the true relationship betwee
>n
>> a monograph and its series. All attributes in the Instance graph shoul
>d
>> solely modify the Instance.
My proposal for a SeriesTitle is to say that an Instance has a Title that refers to a Series resource, modifying the Instance and referring to the Series. The title needs to be a RDF resource in order to do that more on this below.
>
>and cataloguers derive information from the instance to populate
>the (FRBR) work and expression graphs (both starting at an entity of
>class bf:Work)...
>
>When we have resources with several series statement (or a more
>complex series and subseries situation with maybe only one /statement/)
>we should not give up the close connection between the transcribed
>statement(s) (in the "instance graph", including parallel or alternative
>titles if found on the resource) and the corresponding
>links (in the "work graph", including machine-fiendly forms of the
>numbering)
>
>MARC data usually assumes that information coded into (different
>subfields of) the same field is more closely tied to each other
>than to information in different fields, and has the $8 mechanics
>for cases where one field does not suffice. (IIRC for series this
>fails and complementary information may be broken into 490 and 8XX
>without making the connection explicit?)
>
>For my understanding many RDA induced MARC changes tried to
>disentangle those fields which might have mixed information from
>different FRBR levels, thus an AACR2 solution for series information
>might not be suitable for RDA from a design perspective.
>
>Back to Bibframe: When our description is distributed into an "instance
>graph" and a "work graph" (both being distinct named entities)
>then there is one "natural" relation (from instance to its work)
>but for series information (and probably a number of other cases
>if we'd dare to look) it would be desirable to establish a connection
>between a specific subgraph of the instance graph (a "series
>statement") and one or more subgraphs in the work graph (the
>series tracing(s)).
>
>There are probably many technical options to achieve that and my
>impression is that our attempts here in the past week all are in
>the lines of keeping thigns "natural", i.e. if things belong
>closely together put them into a graph of their own (a bf:serialPart?).
>
>We probably should not expect the instance and work graphs to be
>of maximal flatness, i.e.
>
>A a bf:Work;
> any other work attributes.
>
>B a bf:Instance;
> bf:instanceOf A;
> any other instance attributes.
>
>but rather A and B swapping out common subgraphs like those for
>series which mix work and instance attributes like in our series
>examples:
>
><A> a bf:Work;
> ex:swappedOutSerialPartData _C;
> some other work attributes.
>
><B> a bf:Instance;
> bf:instanceOf <A>;
> ex:swappedOutSerialPartData _C;
> some other instance attributes.
>
>_C a bf:serialPart;
> bf:series <Work1> ;
> rdfs:label “Series in English ; Vol. a-3”@en ;
> ex:instanceTitle “Series in English" ;
> ex:caption "Volume" ;
> ex:enumeration "A3" ;
> do ISSNs and other numbers which can directly be taken from the item
>also belong here?
> .
>
>
>Depending on the business rules (will A and B always be transmitted
>together?) these swapped out subgraphs will need to be named (i.e.
>have explicit identifiers of their own) and will have to carry
>explicit links to the bf:Work and bf:Instances they are pertaining
>to.
>
>One might see this as a basic description of the standalone resource
>(with work and instance graph) amended by the data necessesary
>(on work and instance level) to establish the resource as a part
>of a serial / series work.
>
>This additional subgraph could be enriched with any other information
>pertaining to the series, at least as long as this can be taken
>from the (single) volume it belongs to. Thus for some cataloguing
>contexts the actual link to the bf:series might be dispensable
>(but for me it is always a bf:Work to bf:Work relation backed by
>transcriptions from the instance thus even without the bf:series
>statement this swapped out graph is not something which purely
>resides in the "instance graph").
+1
I’ve been mulling over the idea of a subgraph that patterns a Work serving a SeriesPart role. The SeriesTitle on the Instance could refer to that role, and/or the Instance itself could be related as a manifestation of the PartRole. This way (as Mac pointed out) a Work could serve a SeriesPart role with other Series.
Or… we could use bf:Works as Expressions. With a bf:Work as Expression (the Expression serving the “subgraph”) and Titles as RDF resources we could say:
<Work1> a <bf:Series> .
<Work2> a <bf:Text> .
<Work3> bf:expressionOf <Work2> ;
bf:instance <Instance1> ;
bf:series <Work1> ;
ex:seriesPartNumber “3” ;
ex:seriesPartType <ex:VolumeRDFResource>.
<Instance1> bf:title <SeriesTitle1> .
<SeriesTitle1> rdfs:label “Blah Blah Blah Series”@mumble ;
a bf:CoverTitle.
bf:descriptionSource <Instance1> ;
bf:relatedTo <Work1> .
Note, Work2 can be used in contexts outside of the Series, and the SeriesTitle can refer to the Series RDF resource and have additional types. I’ve also made an RDF resource for the concept of volume so that it could be more machine actionable. How the word “volume” appears on the instance would be a literal value in the Instance title graph.
Off to the surf… to erase the fried hard drive.
Yours,
Steven
>
>viele Gruesse
>Thomas Berger
>-----BEGIN PGP SIGNATURE-----
>Version: GnuPG v1
>Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
>
>iJwEAQECAAYFAlXGhVYACgkQYhMlmJ6W47OZCQQAokOkcQgN95C6HzC0j/8oBk6j
>TN3bkZEhf51EfCME5lNpAIRS1kmSum1GUuNiExKpcT93VkDvGMW2fZEqetWlmDoG
>RpYcsiJkljn/SasSZ2pgIRcoHo0qLbG4UD70bdcsOs6kM2y5mp4ZxA9PWzreKAlc
>lFdjXfWQyuFizlyRS3I=
>=XxiK
>-----END PGP SIGNATURE-----
|