When I think of search engines that cluster results, I don’t think of root vegetables. But I like Carrot. I like its clusters. Some of them make me laugh. You may like it too if you try it; it’s at http://demo.carrot2.org/demo-stable/main . (Carrot.org is apparently some anti-spam thing that’s launching in January 2008.)

Clustering search engines, if you don’t remember, gather search engine results into different topics instead of providing them in one huge pile. Searching for chips, for example, might give you topics for chocolate chips, poker chips, computer chips, wood chips, etc.

I searched Carrot for carrot. I got the expected Carrot Cake and Carrot Juice, but also results for Carrot Museum and Carrot Top Records, which is apparently a record label in Chicago. Only 98 results were returned, as Carrot appears to metasearch five search engines and provide the top 20 results from each. A third tab shows you the sources from Carrot’s search, and an additional tab shows the site suffixes that the results are coming from. My carrot search grabbed a surprising number of results from .uk and .gov.

I did find some of Carrot’s clusterings not particularly useful; “Easy” and “News” aren’t much in the way of topics. Carrot however does offer different clustering options. Click on the “Show Search Options” and you’ll get a dropdown menu with six different ways to cluster your search results. “Lingo” appears to be the default, but I liked STC as an option as well (it’s the last item on the drop-down menu.)

Are you less interested in HOW Carrot is clustering than WHAT it’s clustering? In addition to Web searches, Carrot also clusters Yahoo News search, Wikipedia, and PubMed, among others. I am intrigued by the ability to try different clusterings on the fly; I’m going to use this site again.