When Does a Controlled Vocabulary = Keyword Pollution?

[Martin Beebee]

Hi All,

So I'm refining my CV, trying to make it as logical and easy to deal with as possible. One of the areas that I'm struggling with, though, is including enough keywords to maintain a logical (and easy to use) CV without "polluting" the field with keywords that may not be relevant to others. (All of the keywords in my IPTC keywords field are used both by my website and other stock agencies for anonymous searching, so everything that goes in there is (or should be) relevant.)

Here's a perfect example: plant and animals. Organisms are already set up with a hierarchy which would be easy to use in a CV: Family, Genus, Species. So one could imagine the following hierarchy:

--plants
----Cactaceae [family name]
--------Carnegiea gigantea, saguaro [species]
--------Opuntia bigelovii, teddy bear cholla
----Polemoniaceae
--------Phlox diffusa, spreading phlox

So adding the family name aids tremendously in organizing species in the CV, but does including the family name -- which most people would never use for searching -- constitute keyword pollution? Is this unnecessarily obsessing over things that really don't matter much? (I've been known to do that.)

Your thoughts?

Thanks!
Martin

[peterkrogh]

Martin,
It's a very interesting question - maybe best to ask at controlledvocabulary.com

It really all depends on the purpose and the collection.  I think you can make a good case for the full tree of latin names. Most people won't search on it, but those who do will find it helpful. 

It gets trickier with some other kinds of data.  If all your locations are listed as
Earth
>North America
>>Maryland
>>>Kensington
>>>>3301 Oberon St.

Well, North America is a subset of Earth, but that means that every image taken on or of the planet will be tagged Earth. This makes the keyword less valuable. (Of course, Location information should really live in the "Location" fields, but not all software/agencies support this.)

And then there's a related, but separate problem.  How large does a dog need to be in the frame (or how much does the picture need to be "about" the dog) before "dog" gets added to the keywords?

The incentive to the individual contributor - make my pictures more frequently found - can work against the value of the collection as a whole.  Some of this is about policy of the agency/collection, and some of it is about information architecture, including the need for new universal metadata categories that let the user define more about any particular keyword.
Peter

[Chris Bishop]

Following on from PK's reply. I use
Places
  Europe
    Czec Republic
      Prague

This links all the above incl Places, when Prague is selected. However I manually "untick" Places. I can't see any value to searching on Places
Is it sensible to go further? I have two photos of the famous clock and four of Charles' Bridge (for example) How are these identified for future searching? Currently I find Prague then skip through the photos for narrower selections.
This may be a longwinded explanation, but is there "another way" Should it be keyworded to a further level?
Chris Bishop

[peterkrogh]

Chris,
I'd suggest putting Place names in the proper fields - the Location fields.  I discussed this a bit at the end of  this thread:

http://thedambook.com/smf/index.php?topic=2477.0

Peter