Have you ever used Google Image Search trying to find that-funny-thing-you-cannot-remember-the-name?
Have you felt the sense of frustration trying to remember a clue or something else related to that thing, but not having idea about what to write in the box search?
Well, this happen to me very often, and one of the most exciting practical application of Artificial Intelligence is the ability to retrieve the information not by a word match, but by connected topic.
Applied to the WWW it becomes the (in)-famous Semantic Web.

This post is not about what the Semantic Web is, or what it can be useful (a very well-known cousin of Semantic Web is the Semantic Desktop, and one of the most promising technology related is NEPOMUK).
What this is all about is that a piece of real Semantic Web just landed few days ago.
And of course it comes from Mountain View: let’s give a cheer to Google Image Swirl.

Let me introduce Google Image Swirl

Google Image Swirl lets you search for images (just like Google Image Search) but it presents the results in clusters or groups.
Each cluster represented by a thumbnail and represents images with similar appearance and meaning.
By clicking each cluster you can access the group and the results are showed hierarchically in a swirl. Try to narrow you search by clicking one of the thumbnails. The images are reordered and the swirl become like the Debian logo.

A meta-swirl: searching for a swirl within Swirl!

It is still experimental and handles just about 200.000 queries. However, this is a very cool way to look for that-funny-thing-you-cannot-remember-the-name. In the future, when it will become more beta-than-alpha, an image could be researched by grouping the topics and narrowing by simpy looking at the bunch of clusters.

Looking forward for that day…