It's an interesting problem - with no single solution.
I wanted to use SHIC but found it had limited use. I used the NMR thesauri
where appropriate (lots of our images were of buildings, streets etc). I
also looked at the Getty AAT and the Library of Congress.
Like Paul Connell, I set up multiple classifications (called 'themes' and
'subject' in our system) because most pictures fitted more than one
category, and were likely to be of interest to different types of audiences.
I did set up browse categories (based on the 'themes'). As the image
databases get bigger, it does become more difficult to find appropriate
images so I tried to build into the system the ability to carry out more
focused searches that did not just pick up on words that happened to be in
the descriptive text/interpretation accompanying the pictures. Such focused
searches can be refined further at a future date (whenever more money is
found for a further project!). I found it quite important to separate out
the visual content and the semantic content when describing the images.
We also discussed the problem at a regional level - in an attempt to try to
use the same thesauri/word lists where possible. We did come to the
conclusion that more work needs to be done on this issue - particularly
where organisations are trying to serve both the general public and
specialist audiences.
Janet Davis
|