On 7/27/15 7:39 PM, Stephen Hearn
wrote:
[log in to unmask]"
type="cite">
In RDF, you don't "modify" by adding on to a string. You create
graphs of related information, and everything is done with triples.
Relators in RDF won't be added on to the end of string. We had a
long discussion here about relators and how they would be handled,
but there's no "adding a subfield to a string" in RDF. The problem
regarding relators, however, similar to the series problem, in that
there is information that is logically a link to an entity (a
person, or a series), and information that is specific to the
instance being described, such as the role of the person or the
enumerated position of the instance within the series.
Putting the series question into words, instanceX is a member of
seriesY AND instanceX has sequence designation or number "N" within
seriesY. In design terms, the series is identified with URI and
therefore is a linking data element. The series number or sequence
is a string.
In pseudo-code: [*warning, possibly bad code ahead; corrections
welcome]
instanceX bf:seriesStatement [
seriesID <seriesY> ;
enumeration "N" ] .
In triples:
instanceX inSeries _:b7 .
_:b7 seriesID <seriesY> .
_:b7 enumeration "N" .
This looks like what Joe and Theo came up with, with the difference
that the series in this case is a link to a graph of information
that has the data about the series which they include directly in
their series statement. In their case, however, the series
information is repeated in each series statement, which not only
isn't necessary, it might not be a good idea. However, the three
triples above are a bit deceptive because the series information is
linked with a URI and therefore the triples of seriesY logically
exist as part of the same graph:
<instanceX> ex:inSeries _:b7 .
_:b7 ex:seriesID <seriesY> .
_:b7 ex:enumeration "N" .
<seriesY> a bf:Series .
<seriesY> bf:mainTitle âAmerican university studiesâ .
<seriesY> bf:publisher <publisher7> .
<publisher7> a bf:Agent .
<publisher7> a bf:CorporateBody .
<publisher7> ex:prefName "American University" .
In other words, the existence of an identifier in the object
position (the right hand side) of a triple means that all of the
triples with that same identifier in the subject position (the left
hand side) are also part of the same graph. We tend to show
incomplete graphs because they are more concise, but conceptually
graphs are complete. The advantage of this approach is that you can
include all of the information about seriesY in any graph simply by
referring to seriesY as the object of any triple.
If any of the information here isn't available to your system, then
you don't have a complete graph, but, as we do with library
databases today, it is only sensible that all currently linked
triples will be made available to the library application through
some method (copying, caching, etc.). So it's best to think about
all identified entities as being at least virtually local graphs,
and that all information is complete.
kc
[log in to unmask]"
type="cite">
--
Karen Coyle
[log in to unmask] http://kcoyle.net
m: +1-510-435-8234
skype: kcoylenet/+1-510-984-3600