Semi Powertagging
February 8, 2010
Here is the semantic dispersion problem: people who see a picture of a large cat may tag it with Panther, not knowing that this word refers to three different species on three continents. A couch and a sofa are different tags but mean the same thing. A bunker means one thing to a golfer and something else to a soldier. A discussion about Superbowl 45 is the same as a discussion about Superbowl45 or SB45 or Superbowl XLV. Look for Nowali mocassins on eBay and you won’t find any, but look for Nowa Li mocassins, and there they are. So tags are helpful, but they would be even more helpful if people would use the same word to mean the same thing, and if they could set up synonyms to map to the key terms, then searches would become much more accurate. There are many tagging and bookmarking web sites, but they all have this semantic dispersion problem – they don’t try to map different keywords to the same topic.
One tool that does is Semanti.net, a browser plug-in I use to help find sites I’ve already visited. It has an internal mapping system that helps you find web sites you’ve visited, even if you can’t remember the keywords on the site. You try to find them by concept or topic instead. It works better than Delicious, but not a lot. If everyone used Semanti.net, we could all tag the web in a more unified, more useful way, but most people don’t have Semanti.net installed. And Semanti.net uses a proprietary topic map, which then competes with all the other topic maps out there. So how will we eventually get to a common framework for tagging?
The full solution isn’t here yet, but some of the pieces are falling into place.
CommonTag solves the semantic dispersion problem by setting up a public resource containing the most common tags and their concepts. The idea is to tag all the concepts in a document (blog post, report, PDF – anything online) using their common tags, so that stories using the same tag will be about the same thing. It’s lightweight, it’s easy, and it could revolutionize the way we find information. Imagine having a commontag checker, the same way you have a spell checker. It could suggest common tags for you as you write, and all you have to do is say “okay” and the tags go in. As more and more people use the system, the web gets smarter, because the content becomes less ambiguous and easier to find. And it’s not a big rocket-science project either, it’s a project that’s already going and just needs momentum that YOU can provide. It may not be perfectly semantic, but it’s a big win, adding quality structure to free text and making the web that much more semantically enabled. You can imagine other systems and refinements being built on top of it.
Other interesting projects includeOpenVocab.org, where you can try to help define a common vocabulary for pretty much anything, and Microsyntax.org, which is trying to make Twitter more semantic, so software can recognize what people are talking about. For example, Microsyntax.org is working on a set of emergency codes people can use to help disperse information in an emergency, like a tsunami or earth quake. This is great. This is exactly the kind of thing we need, and it helps bridge the world of the formal semantic web with text-based systems like Twitter.
And the topic map we could all use to give meaning to the universal tags is Umbel.org, already developed (and free to use) by none other than Mike Bergman (see below).
So you see? We really are making progress. Don’t let anyone tell you otherwise. Step by step, we’re finding ways to make the semantic web easy. Soon it will become part of our everyday lives without most of us even knowing it. Explore the foundations area of this web site to learn more about what’s coming.
