Re: Definition of "facet"

In message <AAEKLFPLCPPCFCOACDKIMEJDCJAA.aida@acorweb.net> on Fri, 27
Feb 2004, Aida Slavic <aida@acorweb.net> wrote
>
>I don't think anyone care how things are called as far as we have the
>possibility to encode and manage both: top level categories (?facet) and
>subdivisions (?arrays) within those In one of my mails yesterday I tried to
>explain why do we need coding of this structure for managing
>vocabularies

Agreed. I don't mind what they are called so long as we all call them
the same!

>> Both of these approaches lead me to think that it is useful to use the
>> expression "facets" for these groups or ultimate top terms.
>
>I would like to check whether I understood this properly. Do
>you suggest that we call simply FACETS the following?:
>
>       from Stella definition applied in thesauri:
>       Activities/Agents/Objects/Materials/Organisms/Places/Times

Yes, in principle, though any list of this kind is an example and cannot
be definitive.

I go along with the AAT definition
<http://www.getty.edu/research/conducting_research/vocabularies/aat/about
.html> which reads:

"Facets constitute the major subdivisions of the AAT hierarchical
structure. A facet contains a homogeneous class of concepts, the members
of which share characteristics that distinguish them from members of
other classes. For example, the term marble refers to a substance used
in the creation of art and architecture, and it is found as a preferred
term (descriptor) in the Materials facet. The term Impressionist denotes
a visually distinctive style of art, and it is listed as a preferred
term in the Styles and Periods facet.

Facets and Hierarchies in the AAT:

ASSOCIATED CONCEPTS FACET
PHYSICAL ATTRIBUTES FACET
STYLES AND PERIODS FACET
AGENTS FACET
ACTIVITIES FACET
MATERIALS FACET
OBJECTS FACET"

This doesn't say anything about roles.

>       Mills/Broughton applied in BC2:
>Thing/kind/part/property/material/process/operation/patient/product/by-
>product/agent/space/time

I'm not so happy about calling these "facets", because some of them
depend on the role that something has rather than the fundamental
category to which it belongs. This is a rule for citation order, in
which I would say that the "thing" or "material" facets may occur more
than once, as "thing", "material", "patient", "product" and "by-product"
for example. Your other examples below are versions of this, or mixtures
of the "role" and "category" criteria, and thus unsatisfactory in
principle. That's not to say that they don't work in practice, because
humans can live with a certain amount of fuzziness; it's when you try to
tighten this up so that machines can understand it that the anomalies
become evident.

>       Ranganathan applied in CC:
>       personality/matter/energy/space/time

>       Coates,Lloyd, Simandl applied in BSO:
>       tools/operations/processes/parts/objects of study, or product, or
>total system
>
>       Universal Decimal Classification (common concepts + facets):
>       common concepts: processes/properties/materials/persons/ethnic
>grouping/time/space/form/language
>       usual facets within disciplines:
>things/kind/part/property/material/processes/operation/product/agent/
>
>And that division that comes under these should be called ARRAYS???

Yes.

>If so I can see the following
>- advantage
>This terminology will easily be understood by people from bibliographic
>domain... especially here in UK. Classificationists and those building
>thesauri will be at ease with this.
>
>-disadvantage
>a)  This kind of terminology/definition is arbitrary and based on
>assumption that everyone knows that the facet of this kind is based on the
>theory of fundamental/general facet categories
>
>b) as 'facet' is usually defined technically as a result of a division by
>a single criterion ..most of people outside bibliographic world expect
>that content of 'facet' would be a simple list of mutually exclusive
>concepts.

We have problems because people from different backgrounds interpret
terms differently. My point is that even if these terms have been used
with different meanings by respected authorities in the past, that is
not a reason not to try to rationalise the situation now and restrict
the terms to a single, well-defined meaning.

>I think the mail I got from P. Murray illustrate well this last point...
>
><snip>
>Implementations of "faceting" grounded in library science often take the
>form of subdividing a class by characteristics -- for example:
>
>Persons according to sex and Persons according to family or other kinship
>relation (Source: "Facet Analytical Theory for Managing Knowledge Structure
>for Humanities," http://www.ucl.ac.uk/fatks/o_person.htm, 04-sep-2003).

>People by gender and people by occupation (Art & Architecture Thesaurus
>Browser,
>http://www.getty.edu/research/tools/vocabulary/aat/hierarchies.html)
>
>"Both of these examples are described as examples of "faceted" approaches.
>But I find that these examples are inconsistent with my understanding of
>basic principles of faceting and appear not to map well to principles in
>development of computer ontologies.

Indeed. I think that there is a distinction between the definition of
"facet" and the broader concept of "facet analytical theory" or "faceted
approaches"; this broader concept includes the idea of organising arrays
of sibling concepts according to defined characteristics of division.

>From my perspective, Persons according to sex is already a composite term
>(composite subject) -- persons + gender. Same for people by occupation --
>people + occupation. Occupation could clearly be a hierarchical facet by
>itself.

You have to draw a line somewhere in deciding how far to go in factoring
compound concepts into their constituents. The British Standard for
thesaurus construction gives some guidance on this.

>Foskett asserts that an analysis of facets is correct if foci (topics/concepts
>in context within a facet) are mutually exclusive -- "that is, we cannot
>envisage a composite subject which consists of two foci from the same
>facet." (A.C. Foskett, The Subject Approach to Information.) But in one of
>the examples above, it appears that you can create the composite subject
>male cousins.

I think that the point here is that when you set up an array with a
specified characteristic of division, you should try to make the
subdivisions mutually exclusive, so that you cannot combine two foci
from the same array. You cannot have a male female, for example, or an
adult teenager, if the arrays are specified as follows:

people
        <people by gender>
        males
        females

        <people by age>
        children (0-12 years)
        teenagers (13-19 years)
        adults (over 20 years)

I'm not sure that this rule need be enforced absolutely, though, or else
perhaps some refinement or interpretation is needed. In the example I
gave in my previous message, it is possible to have an array of
"materials by function" in which a single material can be both a
lubricant and a foodstuff (animal fat or castor oil, for example).

Also, in the array

        <people by relationship>
        daughters
        sons
        fathers
        mothers
        cousins

someone cannot be related to another single person as both a son and a
father, though a single person can be both a son and a father at the
same time. Care is needed on how you interpret these.

>In my understanding of faceted knowledge organization, inheritance of
>properties is implied in facet hierarchies, but inheritance (especially
>inheritance of all properties) is not formally specified in any examples of
>faceted classification I have seen. In computer "ontologies," however,
>inheritance of properties is usually considered a basic formal requirement:
>"Properties become more useful for knowledge modeling when they are
>specified at a general class level and then inherited consistently by
>subclasses and instances." (Deborah L. McGuinness, "Ontologies Come of Age,"
>p. 177, in  Fensel et al, editors, Spinning the Semantic Web.) What do
>experts like yourself believe about the role of inheritance in facets?

I think that inheritance should apply to all concepts related by the
generic hierarchical ("is-a") relationship, and I think that membership
of a facet is one aspect of this. In my last message I said that one way
of defining a facet is to work up a generic tree until you can go no
higher, and you end up with a facet definition, being the "fundamental
category" to which that tree of concepts belongs.

Leonard

-- 
Willpower Information       (Partners: Dr Leonard D Will, Sheena E Will)
Information Management Consultants              Tel: +44 (0)20 8372 0092
27 Calshot Way, Enfield, Middlesex EN2 7BQ, UK. Fax: +44 (0)870 051 7276
L.Will@Willpowerinfo.co.uk               Sheena.Will@Willpowerinfo.co.uk
---------------- <URL:http://www.willpowerinfo.co.uk/> -----------------

Received on Friday, 27 February 2004 12:16:57 UTC