> --- In TaxoCoP@yahoogroups.com, Renato Alves <renatobalves@> wrote:
>
> Hello,
>
> I'm a master course student of UFRJ University, here in Brazil. I've been
> lurking here for a few days, and this is the first time I ask a question for
> this knowledgeable community.
>
>
>
> I'm researching about Faceted Navigation and Information Retrieval. I've
> been looking over the Internet for some articles/books/white papers about
> which is the best number of facets to use on a classification.

Hello again, Renato

I recently joined this group for a couple of reasons:

One is to find out where the taxonomy community is headed.
Two is to find out whether a taxonomic approach I developed based on natural
language classification will hold up in this community.

I meant to be prococative with my answer to your question, but after reading
your response I realized my mistake. So here is the beginning of a more
definitive answer to your question.

All vocabularies and by extension all languages are referential: the symbols,
strings and sentences that we humans use to communicate are by design one step
removed from 'reality'. Words and pictures are convenient ways to make concepts
more portable.

Taxonomies, in my view, take that concept of portability one step further by
attempting to group 'like' things together. One of the challenges with
taxonomies is that there are almost as many ways of grouping things as there are
people to group them. This phenomenon is especially evident when it comes to
software development. No two analysts will split an identical problem space in
exactly the same way. In almost every company on the planet this leads to the
duplication, segregation and lack of interoperability that causes so many
communication problems in the IT world today.

Another challenge that taxonomies present to practitioners is their inherent
hierarchical nature. There is only one place for any object in a hierarchy, and
that fact often causes a lot of friction. Ask anyone who has attempted to design
a functional classification of documents in an oil and gas company. The
hierarchical nature of taxonomies makes them brittle, which isn't a problem as
long as nothing changes. As we all know language is anything but.

Faceted classification is an attempt to apply what I call intersecting
taxonomies to a grouping exercise. A facet is defined as a mutually exclusive
and exhaustive collection of properties used to classify items in a collection.
So cameras for retail distribution might have brand, price, resolution and
availability as facets. The trick with facets, in addition to the aforementioned
ones, is to make sure they are relevant to the audience and that there are not
so many of them that they overwhelm the people trying to use them, and not so
few that you wind up with too many members in a set.

One other sort of hidden factor is this: if your problem space is relatively
small - like cameras, for example - your chances of satisfying all of the facet
requirements are relatively good. However, if you are charged with creating and
managing facets for all of the objects in your company you could be in a world
of hurt.

I have to cut this a bit short, but if you are interested in where this goes
from here, please let me know and I will continue.

John O'