UW professor 'knows it all' about search engines

A UW professor is constructing a search engine that could change the way people surf the Internet.

Oren Etzioni, an associate professor in the department of computer science and engineering, is constructing a search engine called KnowItAll. The search engine, which is being funded by Google and the National Science Foundation, searches the Web for data and puts it in a list.

"What if you want to find a list of comprehensive things? Say, people who sell mp3 players?" Etzioni said. "Previously, someone would have to run a search on a search engine like Google, and then look at each of the pages to compile that list. KnowItAll extracts information on your behalf, so that you can be interested in broader questions without having to narrow lists."

KnowItAll takes an input noun and finds sentences on Web sites that contain that noun while looking for words that often appear after it. This, according to Etzioni, prevents the user from getting results that would otherwise not make sense.

Etzioni's motivation to create KnowItAll came from two earlier projects, Open Mind and MULDER.

Open Mind is an artificial intelligence project that was created at MIT. For the project, volunteers type sentences into a Web interface to help Open Mind learn.

Etzioni believed that he could come up with a more efficient way to run Open Mind.

"I thought to myself, 'This is silly. We already have billions of sentences on the Web,'" Etzioni said.

MULDER -- a question-and-answer system -- was one of Etzioni's earlier projects. "It was a reverse query system, so you ask it a question like, 'Who killed JFK?' and it responded with: 'JFK was killed by ... blank.' MULDER actually answered 'Elvis' because that was the phrase that most often appeared in its searches," he explains with a chuckle. "KnowItAll is MULDER on steroids and a better way to do Open Mind."

KnowItAll will probably never make it to the public in its current form.

"The code itself can not be used because it was really not written with a professional application in mind," said Etzioni. "The most important thing that comes out of this project is the ideas and the papers."

In William Gibson's 1984 sci-fi novel Neuromancer, he painted a vivid picture of the future of computing. Gibson, an English major with almost no background in computer science, coined terms like "cyberspace" and "the matrix," and described his imaginary worldwide network in remarkable detail.

"In another 20 years, we may actually begin to ask the kind of complex questions that we saw in Neuromancer," said Etzioni. "But to have those technologies in the future means that we must start today."