[thelist] re: developing a system for keywording images in an imagebank

[Tony Light]

--
[ Picked text/plain from multipart/alternative ]
Dunstan wrote:
> Do I just look at each image and write down every word that comes to mind and
> hope that I put in some similar words each time, or do I have a large set of
> keywords which I pick from to ensure some kind of consitancy?

My approach to a very similar problem was to create a small list of what I thought were good useful example keywords.
When a new picture was entered, I showed a text box into which the user could type a list of keywords, comma separated.
By the side of this text box was a scrolling list of existing keywords - with a button (and also doubleclick) to add the selected keywords in the scrolling list to the text box.  So the user could easily mix existing keywords with new ones.
Because my volumes were not huge, I held the keywords with the pictures and just looked for distinct keywords to populate the scrolling list - but if your volume of pictures is large compared to the likely number of keywords, then you might want to keep a separate list of keywords.

Pros:
No brain strain thinking of keywords.
No ticking off for any you missed.
Users get the words they want.

Cons:
Needs Javascript to work well (but even without it, the user can see all previously used keywords)
There will always be people who will not bother looking - particularly if the list gets long. (But you could always revisit the keyword list and rationalise it periodically).

Sounds like you have no problem writing the code - but if you want to see mine, email me.

Cheers,
    Tony.

PS Sorry if this is too late to help - I'm on digest.




--