The FRBR WEMI classes are aspects of a whole; a "full"
catalogue record will usually consist of a Work, an Expression,
a Manifestation, and an Item. A significant utility of FRBR is
that it gives cataloguers a better context for their activity,
as the WEMI classes and individual properties are assigned to
the four identified user tasks (find, identify, select,
acquire/obtain). A corollary is that FRBR makes it easier to
present the metadata in more useful ways to the user;
typically, the user can drill-down from Work (essentially
equivalent to the brief author/title display often seen in
library interfaces) to Expression (content-related stuff like
editions, translations) to Manifestation (carrier-related
stuff) to Item (is it on the shelf?) There is also the
potential for more effective metadata management through
reduction of duplication (one Work record can be linked to
multiple Expression records, etc.) and increased sharing of the
cataloguing effort (only create a Manifestation if no-one else
has done so, and then link it to an existing Expression)
But I see WEMI as an intermediate stage twixt the record and
the statement. The WEMI classes essentially group bibliographic
metadata properties at a lower granularity than the traditional
full record, and perhaps the greatest utility of FRBR will be
to break the hegemony of the catalogue record in the cult(ure)
of cataloguing. See my presentation at
http://gordondunsire.com/pubs/pres/EvolCatRec.ppt
(also listed on the wiki).