[Sigia-l] my first taxonomy ( recieved answers )

[David R. Austen]

Hello Javier, Todd, Adam, Kathryn and all:

What a great thread.

I am personally curious. Not to take anything away from SIGIA--not for
a moment--but is there another e-list dedicated to practical
application of the principals of classification, thesauri and
taxonomy? Is there a website that anybody would recommend? Another
SIG?

Or is SIG-IA the best natural home for these discussions?

The friendly, helpful tone and quality of the information here is just
great. Shall we all continue discussions here on SIG-IA from time to
time?

If there is substantial interest here I would suggest we add some
resources to the SIG-IA Web site. Contributions would be welcome and
all would be credited to the authors.


Cheers,


David Austen




Friday, May 17, 2002, 10:28:01 AM, you wrote:

jv> Thanks a lot for helping.

jv> These are the answeres I recieved:

jv> Todd Levy wrote:

jv>   > Hey there -
jv>   >
jv>   > Since this has been the most interesting question on this mailing list
jv>   > for as long as I can remember, I've come out of a very long lurk.
jv>   >
jv>   > I can't really answer the questions about controlled vocabs, but I do
jv>   > have some insights into the more theoretical stuff below.
jv>   >
jv>   >> a taxonomy consists of a hierarchical tree, where on every terminal
jv>   >> point there is a special type of node where I can attach my keywords
jv>   >> (terms) and associate the content. Each node's label is also a
jv>   >> keyword in itself.
jv>   >
jv>   >
jv>   > 1)- Is this model right?
jv>   >
jv>   > Essentially, yes.
jv>   >
jv>   > Most taxonomy arguments start over the difference between your basic
jv>   > taxonomy, which is just classification according to a pre-determined
jv>   > system, and pure taxonomy, separating elements of a group (taxon) into
jv>   > subgroups (taxa) that are mutually exclusive, unambiguous, and taken
jv>   > together, include all possibilities.
jv>   >
jv>   > Next people argue over hierarchical or not. Some are hierarchical, like
jv>   > a family tree or the animal kingdom (also pure). Others are not
jv>   > hierarchical, like IKEA Stores, the Lands End catalog, and most Web
jv> sites.
jv>   >
jv>   > FWIW, the example below isn't de facto hierarchical as you could
jv>   > accurately present it as...
jv>   >
jv>   >     -Music
jv>   >         -Bands
jv>   >             -Pop
jv>   >             -Rock
jv>   >         -Solo Artists
jv>   >             -Pop
jv>   >             -Rock
jv>   >
jv>   > ... even though I personal prefer your original model.
jv>   >
jv>   >> 2)- Can the higher level nodes also have related
jv>   >> keywords and articles?
jv>   >
jv>   >
jv>   > Yes, although some people do not do a good job of building Web pages to
jv>   > accurately distinguish the drill-down portion of the page from the
jv>   > related content portion.
jv>   >
jv>   >> 3)- Do each of my tree's nodes have to be unique?
jv>   >> Can some structures be repeated in this fashion?:
jv>   >
jv>   >
jv>   >     -Music
jv>   >         -Rock
jv>   >             -Bands
jv>   >             -Solo Artists
jv>   >         -Pop
jv>   >             -Bands
jv>   >             -Solo Artists
jv>   >
jv>   > No, and Sure. The nodes don't HAVE to be anything. And you CAN repeat
jv>   > whatever you'd like. See above re: pure and not pure.
jv>   >
jv>   > I'm not sure if you're familiar with faceted classification, but that
jv>   > may be work exploring for you contents.
jv>   >
jv>   > Music Items
jv>   >
jv>   > * Genres
jv>   >   - Rock | Pop
jv>   >
jv>   > * Performer Types:
jv>   >   - Solo Artists | Bands
jv>   >
jv>   > Definitely check out the Flamenco Image Browser
jv>   > http://bailando.sims.berkeley.edu/flamenco.html




jv> Adam Doerr wrote:

jv> [ snipped some stuff that Adam didn't feel confident enough to share to
jv> the list because it?s also his first attempt, but this comment is
jv> important ]

jv>   > It's unfortunate how few good resources there are in this area, at
jv>   > least that I've found. There are library science resources, but very
jv>   > little info on actually applying the principles. Let me know if you
jv>   > find any good ones - best I've found so far is Thesaurus Construction
jv>   > and Use: A Practical Manual, which basically a library science
jv>   > textbook (expensive, too!), but it does cover some of these systems.
jv>   > Unfortunately, very little applied info.




jv> Kathryn Lewellen wrote:


jv>   > Terminology can get very confusing when talking about
jv>   > taxonomies-for example, taxonomies are controlled
jv>   > vocabularies. Thesauri are controlled vocabularies
jv>   > that leverage synonymous, associative and hierarchical
jv>   > relationships.  Keywords are all together different,
jv>   > at least in my mind.  You bring up a lot of questions
jv>   > that  are difficult to clarify in an email.  You're
jv>   > welcome to call me if your questions aren't answered
jv>   > by other listers (work number is 919 990-9407).
jv>   >
jv>   > here's a post from the archives that may help:
jv>   > http://www.info-arch.org/lists/sigia-l/0111/0075.html




jv> javier velasco wrote: [original question]

jv>  > Hi all:
jv>  >
jv>  > I'm finally having the chance to design a taxonomy, and I've run into
jv>  > some questions along the way. I hope some of you can help me with these
jv>  > questions:
jv>  >
jv>  > If I have understood correctly, a taxonomy consists of a hierarchical
jv>  > tree, where on every terminal point there is a special type of node
jv>  > where I can attach my keywords (terms) and associate the content. Each
jv>  > node's label is also a keyword in itself.
jv>  >
jv>  > 1)- Is this model right?
jv>  >
jv>  > 2)- Can the higher level nodes also have related keywords and articles?
jv>  >
jv>  > 3)- Do each of my tree's nodes have to be unique?
jv>  >
jv>  >     Can some structures be repeated in this fashion?:
jv>  >
jv>  >     -Music
jv>  >         -Rock
jv>  >             -Bands
jv>  >             -Solo Artists
jv>  >         -Pop
jv>  >             -Bands
jv>  >             -Solo Artists
jv>  >
jv>  > The other issue that most confuses me is the way this works next to a
jv>  > Controlled Vocabulary / Thesaurus.
jv>  >
jv>  > For what I've undesrstood so far, these are basically lists of preferred
jv>  > terms (a dictionary with no definitions). A thesaurus has a more robust
jv>  > set of rules to establish relationships between terms.
jv>  >
jv>  > 4)- How do you make these two elements work toghether?
jv>  >
jv>  > Dizzily yours...
jv>  >