American researchers claim to have invented the first Internet search engine tuned to uncover scandal as it unfolds gossip as it’s being dished, and able to monitor the prickly views of an increasingly prominent tribe of “instapundits“: the bloggers.

“Online Search,” the working title for a prototype tool by Accenture Technology Labs, in Palo Alto, Calif., goes beyond mere keyword lookups about products, public figures or trends.

Instead, the search focuses on several thousand influential sources of online news and gossip that have traditionally been less accessible to search algorithms — from chat rooms and bulletin boards, to Usenet groups, fan sites and blogs written by amateur scribes. From those, it identifies hot topics, and monitors people’s positive or negative reaction to The Next New Thing.

Record executives who want to track the “buzz” their artists’ new releases are getting online, politicians looking for instant feedback about a policy announcement or speech, and companies seeking speedier, cheaper ways to conduct post-marketing surveillance after a product launch, are among those expected to benefit from the more tailored search.

It’s a potentially cheaper, more efficient alternative to focus groups or polling. The information is date stamped and refreshed daily, and may be used to chart detailed analyses of competitive issues as they evolve, the inventors say.

“It’s reading the Web well enough to understand what’s being discussed, how analysts or customers perceive your company, how your products and reputation is being compared with others, and analysing your impact in the news,” said Gary Boone, a machine learning expert and project group leader, in an exclusive telephone and e-mail interview with CanWest News Service.

The searches reveal more sophisticated analysis than services like Google Alerts, which automatically sends an e-mail each time a desired keyword appears online.

That method “can tell that you’re being mentioned but it won’t tell you what’s being discussed, or how the online community is talking about a person in relation to others and how that [view] changes.”

“If you’re a political adviser and a scandal appears, you want to correct errors quickly, without inflating the scandal. If you’re a company, you need to gauge when a rumour about your product is large enough that you must respond, [and] shut up the moment it starts to fade,” Boone said.

During the U.S. presidential election, Accenture’s engine trolled political news sites, plotting shifts in voter sentiment between George W. Bush and John Kerry, and accurately gauged the negative impact of the Swift Boat Veterans for Truth on Kerry’s campaign, he added.

In the demonstration, engine searches are refreshed daily; but the system could easily be tweaked to capture near-instant monitoring of stocks, Boone said.

The technology can deduce roughly what a conversation is about, even flag musings about the competition. Despite the effort of numerous research groups to design “semantic” searches patterned after information retrieval mechanisms of the human mind, however, no experimental Internet engine as yet could tally how much money a political candidate raised, or predict the outcome of the next election.

“That’s a level of sophistication that will take decades to achieve,” he said.

© The Vancouver Sun 2005