The Searchers
Smart Business, January 2001
by Lisa Moskowitz and Thomas Claburn
"This is the most disruptive technology to hit the Internet in a long time," developer Gene Kan says. He's talking about Gnutella, a music search service that has big portals on the defensive. Gnutella is now being adapted as a general open-source search tool called gPulp (gnutellang.wego.com).
Paul Hagen, a Forrester Research analyst, concedes that the Napster style of search could threaten portal traffic, but he says search portals will end up benefiting from new technology because "search is still, by and large, bad."
While that's not a point Rajiv Parikh, director of product marketing for AltaVista's business solutions group, would concede, he will confess to fixing what isn't broken. "As a leading search provider," he says, "we're definitely going in and looking at all the technologies possible for us to continue to offer Web-wide search capabilities to our users." As an example, he points to Alta Vista's new Raging Search Unplugged (wml.raging.com), an index of Web pages accessible to users of WAP-compliant wireless phones.
Online searches might be imprecise, but iPhrase (www.iphrase.com) aims to change that. Created by three MIT doctorates, iPhrase uses a combination of natural- language recognition and context sensitivity to pull data from multiple sources within a site to deliver the most accurate results.
"We're trying to effectively do what a customer service representative does when you call the 800 number: react flexibly, and retrieve information from any kind of data sources on the site," says Noam Ben-Ozer, iPhrase CEO and cofounder.
Whether iPhrase can perform in the real world remains to be seen, Forrester's Hagen says. But the technology looked promising enough to Charles Schwab & Co., which plans to roll out iPhrase on its site this year.
San Francisco–based Autonomy (www.autonomy.com) is another company actively improving search capabilities, particularly for the corporate market. The company's software uses pattern recognition technology to analyze a site's content and extract meaning from the text. The theory is that determining the meaning of a user's query provides better results than simply matching keywords.
On a different tack, LexiBot, a new application from BrightPlanet.com (www.brightplanet.com), promises to unearth the 99.97 percent of the Web (by the company's arithmetic) not indexed by old-school search engines. Having mined 20,000 of the top database-driven sites, like the University of California at Berkeley's Digital Library Project, BrightPlanet searches dynamic content directly. Previously, users would need to search for a specialized site, then search again from there to find the right information. Although keyword-based, LexiBot can process natural language queries as well, says Michael Bergman, BrightPlanet chairman.
New search technology keeps the big portal players from shutting the door on would-be competitors, but the real opportunity may be in adding value to existing sites. "As you get vertical splinters, you may take some of the traffic," Hagen says. "But if it starts to take off, then watch one of the portals grab the technology and bundle it in."
