Financial reporter Bambi Francisco reports that business-oriented search technology firm Baynote is trying to mimic the human brain to improve search relevance. Francisco quotes Baynote founder Jack Jia as commenting,

…he and his team tried to mimic the brain structure. Hence, the name. The brain segments searchers into peer groups, based on what they had searched for in the past and which pages they browsed, and the route which they took to find certain pages. The brain then looks at how long users stay on certain pages. The pages on which a user spends the most time is deemed the most relevant and then shown to others in that users’ peer group.

While this method of analysis is intriguing, it sounds more like behavioral analysis in a social web setting than a serious attempt to reverse engineer the human brain. Perhaps something got lost in the translation. Still, the attempt to analyze the behavior of a large number of users to better target search results makes sense, at least in some kinds of environments. By following the navigation steps of real users, Baynote can identify the most desirable content. For example, a financial site might get quite a few searches for “account setup”. Traditional search technology might use a variety of factors to rank pages – keyword content, linkage to the page, position in the site hiearchy, etc. A user might be directed to various pages about account setup, and have to click several links to arrive at the page where one has the information or forms necessary for setting up an account; the Baynote algorithm, presumably, would note that searchers for this term might click several times quickly before arriving at the page about opening account, and direct later searches to the correct page without the intermediate steps.

If you, like Surowiecki, believe in the wisdom of crowds, this could be the search engine for you. Decide for yourself how brainlike it is. (Note that it’s not available on a Web scale like Google, Yahoo, or MSN – rather, it’s designed for businesses and websites.)

email