Re: [tm-pubsubj-comment] XTM vs. XFML and Facet Maps

[Murray Altheim <m.altheim@open.ac.uk>]

Mary Nishikawa wrote:

> Muray,
> 
> What are faceted taxonomies?


Mary,

Faceted Taxonomies or Faceted Classification are related concepts
coming out of the library and information sciences community,
formalized by Shiyali Ramamrita Ranganathan in the 1930's (this
lends weight to the idea that nothing is quite so new as we
tend to think in the computing field). Basically, a "facet" is
similar to TM's concept of a "topic" (though perhaps also to a
topic map facet; in use it's unclear to me which). By definition,
a facet is a "clearly definied, mutually exclusive, and collectively
exhaustive aspects, properties or characteristics of a class or
specific subject."

Classification of subjects can basically take one of three forms:

   1. an enumerated system, whereby each category is given a
      somewhat arbitrary identifier.
   2. each category fits into some sort of predetermined taxonomic
      hierarchy, based on some notion of "god-given" (or otherwise)
      "inherent" structure. (My earlier notion of using the Cyc
      ontology for classification in TM would fit into this idea.)
   3. the "analytico-synthetic" classification, where categories
      are devised by the combination of characteristics, or facets.

There are parts of the Dewey Decimal System that are faceted (such
as its music classification), but Dewey itself is an enumerated
system. I played with Dewey as a topic map, but like any enumerated
classification scheme, the enumeration changes all the time so
there's no sense of a "universal taxon."

The idea with Faceted Classification (an implementation of #3), is
that essentially rather than come up with a single taxon for every
known category, one would develop a set of classes representing
various knowledge domains, which are then broken down into basic
elements which are then related by a set of common attributes or
characteristics called "facets." Ranganathan believed that all
facets could be grouped into one of five groups, which he called
"Personality", "Matter", "Energy", "Space" and "Time." Whether he
got the core categories correct is a bit immaterial, as the idea
of using multiple facets to describe a subject rather than
attempting to fit it into some type of taxonomy, hierarchy,
ontology (whatever) is a powerful concept and probably closer to
how humans think.

Historically it's been mistakes in classification that have caused
us a lot of problems (social, political, etc.), whereas keeping
track of facets entails (to my mind) a much reduced requirement
for subjective judgement, eg., it's easier to describe a category
by its characteristics than to devise a category name, eg., "that
man has brown skin" rather than "that man is of the African race"
(while must still define "brown" we aren't caught in trying to
define "African" or "race"). For many thorny problems, faceted
classification (ie., describing things by their characteristics
rather than assigning them a universal category) offers advantages
over "standardized" taxonomies or ontologies.

Anyway, here's a link to some info on Ranganathan's work:

   http://www.slais.ubc.ca/courses/libr517/winter2000/Group7/facet.htm


> At 11:18 02/09/16 +0100, Murray Altheim wrote:
> 
>> Maybe I've been asleep, but I apparently missed announcement of
>> a markup language named XFML, "based on the topicmap standard."
>> It does seem to blend some of the ideas of topic maps and some
>> of the work going on in the Published Subject TC [hence I've
>> posted this message to both groups, though not cross-posted].
>>
>>   eXchangable Faceted Metadata Language
>>   http://xfml.org/
>>   http://xfml.org/spec/
>>
>> Anyone have any experience with this? My first impression is that
>> it's a bit dubious design, but I've not downloaded and played with
>> the software or spent a lot of time investigating yet -- not clear
>> why development of a new markup language was entirely necessary
>> (perhaps merely subsetting the XTM syntax would have preserved
>> some sense of how they relate), or that XFML is actually an
>> improvement. My interest in it comes from the idea of "faceted
>> taxonomies," which in itself is a valuable concept.
>>
>> From the XFML home page:
>>
>>   "Unlike Topicmaps, XFML is designed to be easy to code for,
>>    and easy to understand. It trades power for ease, yet retains
>>    some extremely powerful [sic].
>>
>>    Publishing XFML is conceptually similar to publishing RSS: you
>>    get many of the same advantages: it is easy, and you get a lot
>>    of incoming links and additional functionality possibilities
>>    that would otherwise be extremely difficult to obtain."
>>
>> Is missing some core features (such as subject identity), yet
>> still claims to preserve the ability to merge topics. I suppose
>> that if it's legitimately a subset of XTM, then it's possible to
>> write some sort of converter.
>>
>> Murray
>>
>> ......................................................................
>> Murray Altheim                  <http://kmi.open.ac.uk/people/murray/>
>> Knowledge Media Institute
>> The Open University, Milton Keynes, Bucks, MK7 6AA, UK

Murray

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

      One of the sad things about corporations is that despite
      their name they have no corpora, no body responsible for
      their actions. They are therefore free to do whatever is
      the will of those who control them, and can transmogrify
      as necessary, like ghosts, to thwart those who might try.
                                                          --ma