Keyword sorting

I've spent some time sorting my keywords recently, after watching the fabulous Luminous Landscape video on organising. It's taken a couple weeks, but then I only have about 1400 keywords and a decent starting point on a structure.

Along the way, I tried doing plants according to the scientific hierarchy with kingdom, family, species, clade etc. It was fascinating work (for me at least), but when I got to "grass" and took a look at the pictures in question, it struck me: Do I really want every picture that just happens to have grass as the background to include Plantae, Angiosperms, Monocots, Commelinids, Poales, and Poaceae in its keywords? Not really. So I scratched that idea and made a much simpler structure - for instance, arctic wolf is now in wolf inside canine inside mammal, and that's it. It's a good lithmus test for hierarchies: Would I want these parent keywords along whenever I use this keyword?

I have a few keywords left over that I'm not sure what do do about, especially the almost 3000 pictures that I have actually managed to tag with Nature.


Found a couple tricks on the way:

When moving keywords around, Lightroom has this funny idea of popping keyword folders open apparently for no good reason. Not just the ones you're moving into, see. Entirely different ones that have nothing to do with the keyword you're moving and that you didn't go anywhere near. It turns out, though, that it only happens when you have some pictures showing in the grid view - like you do after you've clicked the arrow on the right side to check out what a keyword really means. I made sure to keep my Quick Collection empty, and then every time I'd checked out some pictures, I'd Apple-B to go to the quick collection. No more random opening.

LR doesn't allow you to select a number of keywords and delete them all at once. If you right-click after selecting multiple keywords, it will just delete one. You can get around this by moving all but one of the keywords you want to delete into the last one and then deleting that, LR will take all the sub-keywords with it.

It's not a bad idea to put the same keyword in multiple places, as long as you used the keyword entry line that gives drop-down suggestions when entering keywords. In that one, you see not just the keyword, but also what it's inside, so you can pick the right one based on context. Just be careful to not overdo it - for instance, adding color keywords inside animals (black > cat, black > dog etc.) will get very messy very fast.

By prepending a ~ to your keyword folders, you can put them at the bottom of the list. That way, you can always see when you have entered new keywords without getting them into the right place. It might be a good idea to do even before sorting out your year-old list of keywords. By starting at the end of the alphabet, you won't end up having to scroll over a lot of keywords to find the group you where looking for (say, because aadvark, anteater, ape, bird, beetle, and canine all came before feline where you want to drop cat into). I didn't actually do that because I was halfway through before I thought of it.

If there are keywords you just can't figure out where to put, put them aside for a while and go on with other ones. Not only will you get something done instead of frustrating yourself, you might find that subsequent revision of your keyword structure or just more familiarity with it tells you where to place them. Out of 1400 keywords, I have 11 I haven't found a good place for.

Don't try to improve the keywording on the pictures you run across while reorganising, or you'll get nowhere. Only if the same keyword has been used with different meanings, or you've created better hierarchy for something you labelled very generally at first should you change what keywords are on the photos.

If you find a keyword that has since turned into a keyword structure, and you want to sort them further down the structure, you may find that some of your pictures have the more precise keywords already. Say you find a few hundred pictures with USA on them, and in the meanwhile you've added Kansas, California, Idaho and Florida under USA in your World Location keyword folder. Some of them, but not all, have the state already on them. To avoid having to re-keyword unnecessarily, use the Quick Collection as a scratch pad. Select and add all the USA pictures to Quick Collection. Then go through the sub-keywords that have a 'partial match' dash next to them, and for each of those, go to those pictures, select them all and remove them from Quick Collection. Eventually you'll end up with a Quick Collection of just photos that need to be re-keyworded to something more precise. This can also be used later when you find a good way to subdivide a keyword.

Keywording assistant

Ideas for a "keywording assistant" type application: Use keywords of other images not just as completions, but as suggestions.

Screen should contain either several lines of keywords from various input strategies (see below) or one big cloud with combined keywords, emphasizing the ones that are repeated from several strategies.

The user can click to pick, after which that keyword disappears and new ones arrive. Issue: How to avoid that the user clicks just as keywords change? Maybe fade in/fade out, maybe have slow movement from the edge, maybe keywords close to the mouse don't change. It's important that there is no interruption, the user should be able to click on several keywords quickly.

To lessen mouse straing, maybe have the top N keywords have letters or numbers assigned.

There are many possible strategies for selecting relevant keywords. Essentially, they find other pictures and add their keywords. It may be useful to have a way to push keywords out that don't apply, to avoid cluttering.

Strategies:

- number of shared keywords
- MPEG-7 image similarity
- image name similarity (e.g. numbering ranges close to the image in question) Only relevant for user's own pictures
- image time similarity
- image geo-tag similarity
- thesaurus similarity (this doesn't pick pictures, but individual keywords)