dis·am·big·u·ous·i·ty

So, I’ve been actively using del.icio.us for about two weeks now (just long enough to know how to spell it without double-checking).  Being the eternal organizer that I am, I’ve already realized my tags were in need of some, errrr, refactoring.  So, I went for it and the results were: O man – this isn’t easy.

Basically, it reinforced by belief that tags need a context.  Just using a big cloud of tags causes me trouble.  I think the trouble bares itself out when I’m deciding which related tags to use for an item.  For example, if I’m about to tag something as Ubuntu, should I also tag it with Linux?  Isn’t that relationship somehow implied already and thus, following data/object modeling techniques, it shouldn’t be “stored”?

I’ve always believed that the beauty of tags is that we break out of the often unnatural “everything has a hierarchy” mold.   For example, back in college I was the webmaster for my lacrosse team.  I wasted many a brain cycles on deciding whether or not the game summary belonged under the “News” section or the “Games” section.  Looking back, it was quite obviously both and tagging would have solved the problem.

On the other hand, the ugliness of tagging is that we now have a “nothing has a hierarchy” problem.  For example, I tagged an entry about wine (the “emulator“) on Ubuntu with the tag “wine”.   You should see where I’m going with this . . . what happens when I want to tag the cork’d website with “wine”?  Well, that’s an ambiguous tag (and of course, look at the title of this blog to see what I think about that).

Where’s the middle-ground?  Tags need a context!  This is kind of where the del.icio.us bundles go, minus some important features.  For example, if I have a web2.0 bundle that includes aggregators, SaaS, mashups, etc., there will obviously be times when I just want to pull-up all the links in web2.0.  Unfortunately, the bundles don’t work like that.  If they were themselves tags, or at at least treated like them in del.icio.us searches, I think the “tags are related” problem would be solved.  Further more, I’d really like to tag web2.0 with “horriblemoniker” – but that’s for another post.

A different problem is in tagging an item with a location.  If you want “37.80327385185865, -122.25723266601562″ to mean anything, you’d have to know that tag is a type of location of the longitude, latitude variety.  This is the same problem with the aforementioned “wine” tag: some tags aren’t nearly as useful if you don’t know their type.  Do tags need namespaces?  Carrying that argument to its logical extent, can some tag namespaces have exclusive owners?

Upon writing this, I’m realizing this is an age-old problem of sorts.  I’ve seen it first hand in attempts at “dotted-line reporting” in the context of business organizations or “multiple-inheritance” in the OOAD context.  I would bet this is quite an issue in the context of biological sciences.

So, who’s solving this problem?  I’m sure there are a ton of resources out there that have waxed poetic about this, so please show me the light.