Skip to Main Content (press enter)



Clusters. This month’s small-to-middling topic on the search horizon. Google has been showing off it’s clustering technologies at the recent Web2.0 conference (Danny Sullivan’s write up and an MP3 of the session) and Vivisimo has taken its search grouping to the next level by launching the somewhat tweely-named Clusty search site.

Clustering is based on Bayesian (A method of combining the likelihood ratio with additional information to produce an overall estimate of the strength of a piece of evidence, named after the Reverent Bayes) principles and is particularly useful when search terms can have source material across completely different areas of interest, as it seeks to group the results in their most likely overlaps first. I am not convinced, however, that clustering will make a significant difference the hunt for spam pages.

Filed in Blog, October 19th, 2004. Leave a comment, or trackback from your own site. Follow comments via RSS

Leave a comment