In other news

[gement]

My job is going so spectacularly well that I don't even feel inclined to take time away to post the really wonderful bits here, but some highlights for my personal benefit (enjoy what you can of them):

Guerrila librarianship (imposing my political views on the controlled vocabulary, mwahahahahah!).
Thinking of 15 distinct uses for the word "queen" or "queens"... then figuring out how, by caption searches, to distinguish between them reliably.
Giggling at the hilarity when we find something that's been automatically categorized All Wrong (the Watergate affair under Infidelity, Flushing NY under Toilet).
Finding pictures of Ascot hats, Chinese ballet, actors I like, first ladies of tiny nations who have really frightening hair, Terry Gilliam sticking his head into one of his new hollow statues, the national Air Guitar champion, London's giant mechanical elephant, Eurovision's new winner who's made up to look like an orc, and Pricasso.
Outwitting the fashionista (who put 15 keywords on one image to signify that an actress is wearing shoes).

And now I really must get back to it.

Comments

[gement.livejournal.com]

... kind of. In that particular case, there's a little note in the cleaning instructions (since this word must be cleaned by hand) that says "watch out for overlap". Whether people listen or not is up to human error.

While that makes my fingernails itch, even I can see that keeping track of repetitions (how often a concept comes up in one caption) would cause more problems than solve them. We'd end up in exactly the same place, with the cleaner going "It's all in Flushing Stadium! I don't care if it says it's in there 5 times! It's done! Leave it!" Particularly with place names, when the caption starts with QUEENS, NY and then goes on to talk about what's happening, you get repetitions automatically; they become more noise.