Tagging Tagging Tagging

January, 2007

I haven’t written anything substantive in months. Work, the holidays and so forth have kept me plenty busy. For a while now, I’ve been meaning to blog some of my thoughts on tagging. I’ve been gardening my tags for over two years now. I’ve watched others use del.icio.us and experimented with my own techniques. Two techniques I’ve tried are pre-coordinating and post-coordinating tags, which I describe briefly below.

Pre-coordinate tags

Pre-coordinate tags are pretty scarce. They are useful within a user’s own collection, but I have not seen any practical application for them socially. Let me explain. A precoordinate tag connects more than one concept or phrase into a single item, like taking apple and osx to make apple.osx. Done correctly, precoordinate tags can provide very rich semantic meaning, like a taxonomy. They can also mitigate one of the biggest weaknesses of tagging - homonyns (words used in more than one way). Unfortunately, pre-coordinate tags don’t work as well as you’d expect. First of all, the very high granularity comes at the price of any connection with other users. If you use a tag like personal_computer.manufacturers.apple there is a lot less chance other people will hit on the exact same tag. A clean, simple tag like apple enjoys hundreds of thousands of happy users, so there is a ready-made community for it. But of course the poor folks that tag a local orchard’s website are drowned among all the computer geeks. The second limitation of pre-coordinate tags is that even though you can construct your own hierarchy with tags, the tagging software doesn’t really work that way. For example food.fruit.apple has no relationship with food or food.fruit. And the third limitation is that there are no emerging standards on how to actually accomplish pre-coordinate tags, probably because they don’t lend themselves to social interaction. In short, I gave up on using these.
Example:
http://del.icio.us/tag/platform:windows

Post-coordinate tags (a.k.a. facets)

Using facets in a tagging environment is simple and powerful, but it requires a fair amount of discipline. For many folks, this defeats the whole point of an easy breezy tagging app. Well, let me first explain how it works and you can decide for yourself. In any faceted classification system, several mutually-exclusive controlled vocabularies are used in parallel. Each item is classified using more than one vocabulary. For instance, a work of art could be classified with the name of the artist, the name of the place it was created, the name of the artistic style or genre it conforms to. Each name would come from a different vocabulary. Other pieces of art would share those names too - the same artist, the same place, the same genre. The names are short and simple enough that anyone can tag with them consistently. And then others can combine the tags in really powerful ways.

But, as I said, it comes at a price. You have to use consistent vocabularies. That sounds easy enough, but it’s tricky in practice since you have to remember which vocabularies to use off the top of your head. Tools like del.icio.us support this practice a bit by recommending tags you’ve used before. Personally, I try to assign everything at least an author, genre, topic, and doctype.

I’ve created a few facet combinations that I use regularly, which I call ‘post-coordinate filters‘. I can subscribe to a really tight set of bookmarks using the RSS, which is vastly better than any single tag. And if I use common tags, it gives people hooks into my tags.

Examples:

Filed under Findability |

Comments are closed.

Recent Bookmarks

All Bookmarks