Comments on: Classifying Web Search Results http://www.adaptivepath.com/blog/2006/09/15/classifying-web-search-results/ Adaptive Path Blog Fri, 16 May 2008 06:24:19 +0000 http://wordpress.org/?v=2.3.3 By: Nils T. Devine http://www.adaptivepath.com/blog/2006/09/15/classifying-web-search-results/#comment-176256 Nils T. Devine Thu, 07 Feb 2008 00:39:03 +0000 http://www.adaptivepath.com/blog/2006/09/15/classifying-web-search-results/#comment-176256 Here's a search utility that I saw recently whose entire purpose is to classify your search query. It's for web searches, but the line of thinking is similar to what you were talking about. What it does is determine the topical site that is most appropriate for your search, and push the query to that engine. It uses Amazon for books, Google Maps for directions/locations, Flickr for photos, and so on. <a href="http://boom.baekdal.com/" rel="nofollow">Boom! Intelligent Search Assistant</a> (<a href="http://www.baekdal.com/future/boom-intelligent-search-assistant/" title="blog entry" rel="nofollow">about</a>). One thing that I found happening after a couple test searches is that if I didn't get the site I expected I would hit back and adjust my query. Given that I'm a total geek, I would quickly learn the correct query keywords to get the site I want, much like using Google's advanced search <a href="http://www.google.com/intl/en/help/refinesearch.html" rel="nofollow">operators</a>. This is acceptable for personal use by a more technical audience (which appears to be how this tool came into being), but needs refinement for a general audience. An internal application of this concept could allow the searcher to correct the engine's guess and pick a more appropriate or specific category. This could avoid the "hit the back button and try again" loop. I'm still skeptical about using this technique on a general audience site, it's probably too strong a filter of the results. Here’s a search utility that I saw recently whose entire purpose is to classify your search query. It’s for web searches, but the line of thinking is similar to what you were talking about. What it does is determine the topical site that is most appropriate for your search, and push the query to that engine. It uses Amazon for books, Google Maps for directions/locations, Flickr for photos, and so on.

Boom! Intelligent Search Assistant (about).

One thing that I found happening after a couple test searches is that if I didn’t get the site I expected I would hit back and adjust my query. Given that I’m a total geek, I would quickly learn the correct query keywords to get the site I want, much like using Google’s advanced search operators. This is acceptable for personal use by a more technical audience (which appears to be how this tool came into being), but needs refinement for a general audience.

An internal application of this concept could allow the searcher to correct the engine’s guess and pick a more appropriate or specific category. This could avoid the “hit the back button and try again” loop. I’m still skeptical about using this technique on a general audience site, it’s probably too strong a filter of the results.

]]>