>    Date: Wed, 18 Dec 2002 14:51:00 -0500 (EST)
>    From: Kathryn La Barre <klabarre <at> indiana.edu>
> Subject: Re: Poor FC example?
...

> (p.12-13)
> What is facet analysis?
> 
> "The sorting of terms in a given field into homogeneous, mutually
> exclusive facets, each derived from the parent universe by a single
> characteristic of division." 

This is one that bothers me.  In theory, it sounds like a good idea to have facets be mutually exclusive, but in
practice, this seems like an unreasonable goal.  For example, what if the wine in question comes from two
countries?  What about cars that are assembled in North America from parts manufactured in Japan?  Or a tax
regulation from the state of New York on the treatment of income from New Jersey, and how that affects a
federal tax return?  For these examples, the most understandable "facets" would be states or countries, I
think, but the exclusivity constraint would be violated.

Just as one of the first rules for real database modeling is that unique identifiers often aren't (e.g ISBNs
are often reused for different books, or a large business may end up with several D&B numbers per business
location because of consolidations and acquisitions), the mutually exclusive rule may not be
attainable for real classification schemes.

NB: For an interesting study of how a categorization scheme evolves and coexists with earlier versions,
check out Dourish et al.:

http://www.ics.uci.edu/~jpd/publications/macadam.html

Ken Meltsner

To unsubscribe from this group, send an email to:
facetedclassification-unsubscribe <at> yahoogroups.com

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/