Surprise, surprise. Google has done something cool.

Google has found a way to not only make the semantic Web a tangible concept for users, but also increase the power of its own semantic engine in the process.

And guess what? It’s actually pretty fun.

The semantic Web, which I discussed in a previous column (Prepare yourself for the next wave of Internet technology: cloud computing) is a step toward making search and Web functions respond in a way that emulates human thought. Anyone who has visited Google’s image search site lately may have seen a small text link asking them to participate in something called ‘Google Image Labeler.’ Those who have participated have been effectively teaching Google’s computers themselves and are in turn adding to the power of the semantic Web.

So what exactly is a strong semantic engine? To put it plainly, it’s when Google’s search appears to return more specific results than you would expect from normal keywords. Let’s say, for example, you Google image search ‘love.’

Right now, you’d get back a lot of paintings and art with the word ‘love’ in the title, a couple o graphics that actually have the word ‘love’ incorporated, a few cute images of elephants making hearts with their trunks and of course, some porn.

With a semantic search, your results might return some of the following: a couple holding hands, a father or mother and their child, the cliche black-and-white image of a young boy giving flowers to a young girl – you know, the stuff that actually comes to mind when you think of love.

The way Google is making this work is astounding in its simplicity. Go to Google’s image search and click on the Image Labeler link: (http://images.google.com/imagelabeler/).

It’s OK, I’ll wait.

OK, there? Good.

Once you’re there, you get to play a game. You’re teamed up with a random partner out in cyberspace who is also playing with Image Labeler. You both get shown the same series of images and must type in labels that describe each photo – quick, you only have two minutes! Some words are off-limits if they are already keywords for that image, and you and your partner get points for each time you write the same label for an image.

When your time is up, you get your final score, see how your partner did and see each image in its context on a Web page, if you choose. If you haven’t guessed by now, these aren’t just random images; they’re images online already, already searchable by Google that you are adding keywords to.

So, you may still be wondering why (ITALICS)you(ITALICS) are adding these keywords. Well, because no one else can. Not everyone adds keywords to their photos online. Even if they do, art is subjective and it could mean something different to someone else. A computer, which can do color recognition well, and face and scene recognition to an extent, can’t see the intangibles that humans recognize, like the love example from before.

Google is pioneering semantic search by putting it in the hands of users who want better results, and, to a lesser extent, those who are just bored. Google’s plan is simple and brilliant: Playing a two-minute word association game is easy and fun, and while one round doesn’t make much of a difference, thousands of people playing the world over does – not to mention the fact that playing and accumulating points is actually fun!

So next time you’re bored in class, instead of firing up solitaire, minesweeper or YouTube, go to Google and teach Google a thing or two.

AJ Chavar is editor in chief of The Daily Orange, but still has trouble with deadlines. His columns try to appear weekly.

Published on November 5, 2007 at 12:00 pm