Spiteri's Simplied Model for Facet Analysis (RESOURCES)

I was wandering through PeterV's blog (http://www.poorbuthappy.com/ease) of
18-feb-2003 and came across his reference to Dr. Louise Spiteri's paper, "A
Simplified Model for Facet Analysis" -- which is now available at
http://aifia.org/pg/a_simplified_model_for_facet_analysis.php.

I recommend this paper highly. Thanks, Peter! (And thanks to Dr. Spiteri, of
course!) Dr. Spiteri does a great service in summarizing key aspects of FC
and highlighting major differences between Ranganathan and the work of the
CRG (Classification Research Group).

For me, reading the paper brought to light several specific areas in which
my expectations of faceted classification differ significantly from
"traditional" FC as described by Dr. Spiteri. This is not to suggest that
the principles described in this excellent paper are "wrong" (especially for
general collections of documents). However, from the perspective of my
interest in organizing the knowledge or organizations (businesses,
associations, government agencies), some of the principles are misleading or
inappropriate. And the set of principles may be incomplete for the
requirements of many organizations.

A few of points that I feel compelled to comment on:


Exhaustiveness
--------------

Spiteri notes:

"[Ranganathan's] Canon of Exhaustiveness states that all classes and
sub-classes in a classification system should present all aspects of their
parent universe."

Later, Spiteri notes: " the CRG believes that no one list may be necessarily
exhaustive or applicable to all subjects. "

Spiteri clearly comes down on the side of the CRG on this point, but it's
worth re-emphasizing that exhaustiveness of categorization is not possible
in rapidly changing organizational knowledgebases. It's far more important
to categorize existing resources thoughtfully but as quickly as possible
than to provide exhaustiveness. For me, FC allows you to do just that.


The "function" example
----------------------

Spiteri provides an example of classification of dogs that includes the
facet FUNCTION.

This, IMHO, is a very bad example. FUNCTION is a vital characteristic when
looking for information about dogs ... or about computer products. In
industry, functionality is often the characteristic of interest to
information-seekers.

But from my point of view, a FUNCTION cannot be a facet. A FUNCTION is
described by *combining* terms from several facets. For example, in the
world of Web caching, the function "Edit metadata" might draw from an ACTION
facet and a DOMAIN OBJECT facet. Note that I am expressing the function as a
natural language statement or declaration. The terms are combined with a
simple syntax.

I don't think it's possible (or desirable) to organize FUNCTIONS into a
single hierarchy that represents a characteristic that is orthogonal to all
other characteristics (expressed as facets). In any case, that may not have
been Spiteri's intent. The discussion of FUNCTION as a facet was brief.

In an ideal online schema, information seekers should be able to traverse
links from terms within a facet to complex terms (e.g., FUNCTIONS)
containing that and other terms  ... and vice versa.



The "cats" example
------------------

The CATS example violates my basic understanding of how facets should be
used and why they are so useful. Spiteri's example appears to integrate
multiple unique characteristics into a single facet. The example looks much
more like a traditional (eg., LLC-style schema). Perhaps I am
misunderstanding her intent or the context of her example.

The facets I would choose might include:

--  BREED (but BREED has its own problems as a facet)
--  PHYSICAL CHARACTERISTICS
         -- coat length
         -- body shape
         -- adult size
         -- longevity
-- ILLNESSES
-- DIET
-- BEHAVIOR
-- HABITAT

"Breed" of dog or cat is not necessarily "permanent." The cat fanciers
association, for example, "declassified" our beloved Himalayans, making them
a subclass of Persians. And it seems that new breeds of cats and dogs are
cropping up all the time.



Synonyms
----------

"Synonym" is a dangerous word. In thesaural relationships, this is often a
"USE FOR" term that points to a "USE" (preferred) term.

In my view, you need to distinguish between aliases (terms that have
*exactly* the same meaning -- usually by virtue of being an alternate
spelling or abbreviation) and synonyms (terms that are very close in
meaning).


Systems of Notation
-------------------

Notation does not generate uniqueness for a term. However, for the sake of
clarity and manageability, every term should have a unique ID -- one that is
persistent, parsable, language-independent, and processable by computer.
(For example, such a unique ID should not begin with a numeral and should
not contain characters that are routinely excluded from use in file names,
URLs, etc.) Uniqueness of a term is defined semantically by its position in
a facet hierarchy and any additional relationships (typed or untyped)
posited between the term and other terms (in other facet hierarchies). This
is also related to the Canon of Mnemonics. The computerization of a schema
should do away with any reliance on mnemonics. All mnemonic schemes for
large collections ultimately break down. Sometimes they break down almost
immediately.


In general, the immediate needs of members of an organization may be so
fundamentally different from the needs of the public from a general
collection that it is unfair to expect all canonical principles of the
faceted approach to classification to apply within organizations.

Maybe what I'm interested in should not be called "faceted classification"
at all??? I hope you all don't decide to kick me out the club. ;-)


     Phil


-------------------------------------------
"I have made this letter longer than usual, only because I have
not had the time to make it shorter."
-- Blaise Pascal, Lettres Provinciales, 1657

Phil Murray -- Chief Knowledge Architect
The Knowledge Management Connection | http://www.KMconnection.com
401-247-7899



----------
Thanks for playing. To unsubscribe from this group, send an email to:
[email protected]

 

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