Faceting

 I wanted to throw something out for consideration. I'm working on a
paper related to FRBR and faceting (faceting secondarily, though). This
is how the picture looks to me at this point. There seem to be two
extremes: FAST seems to be a way to facet subject headings to make
cataloging less expensive, while proposals like those of James Anderson
and Melissa Hoffman (C&CQ, v. 43 no. 1 [2006]) and Pino Buizza and Mauro
Guerrini (C&CQ, v. 34 no. 4 [2002]) seem to be proposals to make
cataloging more expensive. (I'll admit that's not their intention, but
it seems it would be the effect!) 

 

Of course what they have in common is an antipathy to precoordinate,
enumerative headings. They want you to be able to approach every piece
of subject vocabulary independently rather than through "main headings"
and browse screens. In the case of FAST, the indexing is designed for
keyword searching, whereas my best guess about the other two proposals
is that they're a reworking of subject searching to make it more
"transparent." I can hardly believe that the Anderson-Hoffman or
Buizza-Guerrini proposals could be implemented without increasing the
cost of cataloging. But maybe they think that would be compensated for
by lower reference costs. 

 

The point I want to raise is that precoordinate headings and subject
browse screens represent a sort of compromise: they're designed to make
the various distinctions and varieties in subject matter clear enough
without making the subject system completely transparent. For example,
in terms of Anderson-Hoffman's objections, for the book Sex and
conquest: gendered violence, political order, and the European conquest
of the Americas, the LCSH "Dominance (Psychology)-Spain-History" doesn't
make it clear whether the Spanish are doing the dominating (Agent) or
are being dominated (Client). But I wonder if the costs of making that
point transparent would be prohibitive. And is it really worth it?

 

In contrast, FAST makes subject browse screens basically useless. That
is even true to a lesser degree of the "faceted" organization of MeSH
that is used in NLM"s Locator Plus. Not quite as bad as FAST, since
subfield x's are kept with main headings, but you still get things like
"Congresses" with a result set of 83,548. So money has been saved in
cataloging, but the research costs for the user have been increased. 

 

I'm open to the possibility that this "dilemma" is exaggerated. Maybe
you can facet without creating either of those problems. But that's the
way it looks to me at this point, with my limited knowledge.  

            --Ted Gemberling