Can Search Be Made More Sensible?
from the nice-in-theory... dept
With so much attention being placed on search these days, many are starting to explore whether there’s a “next generation” style of search that goes beyond today’s search systems found on the popular engines. Here’s an article suggesting that maybe the next big thing in search is not a search engine but a “sense engine.” Of course, reading the article, you realize this is basically marketing-speak for a clustering engine — such as the one that Vivisimo already offers. Basically, the real end goal is to have a search engine that can figure out exactly what you really want — but in the meantime, all they can do is clump similar findings together and have you pick a group. The technology is still very early stage, and so far, it seems like being smarter about query creation on traditional search engines (basically, ad hoc clustering by adding and/or excluding certain words) is more effective and efficient than using a clustering engine.
Comments on “Can Search Be Made More Sensible?”
Look at Northern Light, it had clustering long ago
Search results were grouped by “folders” which were basically clusters. As of now, northernlight.com doesn’t offer search services to the public (you can get a free trial to check it out), but certainly folders of search results were well-received by Northern Light users.
The clustering itself was done by a bunch of librarians well-trained in taxonomy. They were much more effective than any computer-driven algorithm.
Re: Look at Northern Light, it had clustering long
I don’t know if any one remembers. There was this engine called the Hubat (http://www.hubat.com) that aimed to automatically categorize all the Web that it spiders. It was awfully buggy, but extremely promising.
By the way, instead of clustering search results, it attempted to create a directory of the Web