Use and abuse of occurrence RE: [topicmapmail] Are FacetsReally Simple After All?

[Murray Altheim]

Lars Marius Garshol wrote:
> * Lars Marius Garshol
> |
> | ..you have topic maps you can do *better* than faceted
> | classification, and that you no longer need ....
> 
> * Mark Brand
> | 
> | i am just wondering what is *better* and can you give me an small
> | example represented in FC and the alternative representation.
> 
> The only thing I have to hand that comes even close to this is this
> thread I participated in on SIGIA-L a while back:
> 
>   <URL: http://www.info-arch.org/lists/sigia-l/0301/0459.html >

In that message you wrote:
 >
 > Faceted metadata essentially means replacing a single hierarchy by
 > multiple hierarchy, with one hierarchy for each facet. Now,
 > structurally that's actually the same as having a single hierarchy.
 > (Just add a single node at the top linking all the facets and you've
 > got it.)

Without quoting the entire message, I don't think I've quoted you
out of context.

You don't in this message indicate you've grasped much of anything
about why Faceted Classification is an important metaphor in
information science, and why it provides much of the backbone for
library classification schemes.

For example:
 > [...]
 > And hierarchies fail on several counts in this regard:
 >
 > - you cannot type relationships, so there is only a single kind of
 >   relationship between the nodes,

No, every single facet is a relationship type. There may be a facet
hierarchy, but the relationship between the facet and the entity is
typed by the facet relationship. For example, a book can have an
author facet (perhaps called "hasAuthor"). That's not a hierarchical
relation. In Dublin Core, you have quite a number of relations, such
as "replaces", "isReplacedBy", "hasVersion", "isPartOf", etc. that
have nothing to do with hierarchy. Most FC relations are affinitive
or associative, others are "syntactic", "semantic", etc. In fact,
hierarchical is seldom mentioned. Whether or not the facet hierarchies
are rooted is immaterial.

 > - you cannot type the nodes, so a machine can't tell countries from
 >   diseases from people from animal species,

I have no idea where you got this erroneous notion. Who created a rule
that says you can't "type a node"? There are acts of characterization
(which is a form of typing) all over the place in FC. DCMI has an entire
type vocabulary. Those types are precisely for disambiguation of
resource by country (e.g., from "http://purl.org/dc/terms/RFC1766"),
people (as by creator, or even by educational level, as in
"http://purl.org/dc/terms/educationLevel"). And a more generalized FC
system could certainly classify by disease, or whatever you like.

 > - you cannot assign properties to the nodes beyond one or two fixed
 >   kinds of names and perhaps some untyped URIs to content, and

Nothing in FC states this. There are dozens and dozens of assignable
properties in Dublin Core alone, and it's not a very complicated FC system.

 > - your relationships must form a tree.

No, perhaps there may be a hierarchy for a single facet (though not
necessarily so), but the entire lattice of an FC system definitely
is a complex graph, and no particular part of the graph is any more
important hierarchically than any other -- you approach the graph
from the perspective of a specific query.

 > Obviously this doesn't give you much power to model the real world,
 > and equally obviously facets do not solve any of these problems. (You
 > could theoretically solve the second problem with them, but I doubt
 > that would work well in practice.)

I would have to disagree on "obviously", and I am currently using
FC to solve a real world problem. In fact, I believe that FC is
very similar conceptually to frame systems (facets and their role
in the system are not unlike slots), from which a huge number of
successful knowledge-based systems are constructed or derived.
The facets become the primary means of navigating the content,
where you can obtain all entities having a specific facet value,
such as "author:Durrell, Lawrence", "edition: 1st", or in a more
generalized system (such as Ceryle), "eyecolor: blue",
"birthplace: New York", etc.  FC becomes the primary metaphor/means
of organization, searching and navigation. This fits nicely as a
layer on a Topic Map system. This is not conjecture: I'm doing it.

For more information on FC, and especially FC in use in KBS, check

     Faceted Analytical Theory  (has biblio refs)
     http://www.ucl.ac.uk/fatks/fat.htm

and much of Uta Priss' work is focused on FC, associative structures,
Formal Concept Analysis (FCA), and the like. Her husband, L. John Old,
recently published his dissertation on

   "The semantic structure of Roget's, a Whole Language Thesaurus
   http://www.dcs.napier.ac.uk/~cs171/LJOld/diss/Dissertation.pdf (3.4MB)

which isn't precisely FC, but is all about facets in language and
how classification (which he calls "facets") occurs in Roget's.
Additionally,

    Uta Priss' home page
    http://www.upriss.org.uk/top/research.html

   "Faceted Knowledge Representation"
    http://www.ep.liu.se/ej/etai/2000/002/

   "Description Logic and Faceted Knowledge Representation"
    http://www.upriss.org.uk/papers/dl99.pdf

   "An Algebraic Approach for Specifying Compound Terms in
    Faceted Taxonomies", Tzitzikas et al.
    http://www.dcs.bbk.ac.uk/selene/reports/Facets-KDE.pdf

   "A Hierarchy-Aware Approach to Faceted Classification of
    Object-Oriented Components", Damiani, Fugini, Belletini.
    ACM 1049-331X/99/1000-0425

   "Facet analytical theory as a basis for a knowledge
    organization tool in a subject portal", Broughton, V.
    http://www.ucl.ac.uk/fatks/paper2.htm

e.g., read the first paragraph of:

   "The key role of classification and indexing in view-based
    searching", A. Steven Pollitt.
    http://www.view-based-systems.com/papers/pollifla.htm

et cetera.

Murray

......................................................................
Murray Altheim                    http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK               .

       Subject of Spam:  "Grodepartmentw your peducklingnis"
       (this one took me a minute to decipher)