Source: William Hsu, 785-532-7905, bhsu@k-state.edu
Pronouncer: Hsu is shoe.
News tip prepared by: Jennifer Torline, 785-532-0847, jtorline@k-state.edu

Monday, Feb. 14, 2011

NEWS TIP: WHAT IS 'WATSON?' EXPERT CAN EXPLAIN THE WORKINGS BEHIND IBM'S 'JEOPARDY!' CHALLENGER

MANHATTAN -- A Kansas State University expert on machine learning and artificial intelligence says the IBM-designed Watson will be a strong candidate against "Jeopardy!" champions Ken Jennings and Brad Rutter.

Research by William Hsu, associate professor of computing and information sciences at K-State, involves building data models that help with making decisions and answering questions -- an area that closely relates to the inner workings behind Watson.

Hsu said Watson acquires knowledge in two ways: through questions that have come from previous episodes of "Jeopardy!" and through a Web crawler that collects information from the Internet.

Unlike Deep Blue, an IBM supercomputer that defeated world champion chess player Garry Kasparov in 1997, Watson is able to learn. While Deep Blue won by brute force using some of the most powerful computer hardware of its time, Watson consists of software that can run on less specialized computers. Using mathematical algorithms that are programmed into it, Watson is able to educate itself by crawling the Web and analyzing the resulting data.

Hsu's wager: In its current state, Watson has a better than even chance of winning "Jeopardy!"

Hsu can be reached at 785-532-7905 or bhsu@k-state.edu.