I was recently looking around for ontologies that might be used to
express/exchange information about scholarly articles and the different
versions produced by publishers and held in repositories (like arXiv.org
that I'm involved with). BIBFRAME was appealing because I think that
"FRBR-lite" Work/Instance model of BIBFRAME is a practical and useful
level of granularity. I'm also interested in RDF expressions of library
data as we are creating a new discovery system in our library where all
our MARC records are being mapped into RDF in order to merge them with
additional information held in systems other than our LMS.
I've been following with interest the discussion of annotation in
BIBFRAME and am left wondering what special value is added by BIBFRAME
adopting a model not-quite-compatible with OAC. From [1] it seems that
the BIBFRAME "special sauce" regarding annotations is twofold: first is
a specialized set of types of annotation tied to community needs, and
second is a simple syntax/structure for the use of literals. I don't see
that either of these is a compelling reason for a different approach.
The specialized set of annotation types very nicely maps onto the (more
readily extensible) oa:motivatedBy model where instances of
oa:Motivation could be usefully subclassed. The use of literal bodies
can be handled with the (admittedly slightly more cumbersome)
oa:ContentAsText mechanism. A possible good side-effect is that this
might discourage the use of "not on the web" literal annotations except
in cases such as those Karen Coyle mentioned where user tags perhaps
"deserve to be literal strings, dangling off the edge of information space".
My conclusion is that the value-add of BIBFRAME is not in its annotation
model and I think it would be better to promote a profiled use of OAC to
support interchange within the target community (while also supporting
uses by others not yet imagined, presumably one of the goals of going
the RDF route).
2c,
Simeon
[1] http://bibframe.org/documentation/annotations/20130430.html
|