Okay, IBM's Watson got a little flummoxed about airport names at Chicago versus Toronto, but chalk it up to nervousness at his first live TV appearance. Overall, however, the supercomputer - running on 2,880 POWER7 cores in a cluster of 90 IBM Power 750 servers on Linux - won handily over Jeopardy's all-time reigning human champions, proving the computers have the capability to learn and respond to natural language queries steeped in culture, subtleties and wordplays.
The Watson system has roughly 200 million pages of natural language content - equivalent to reading 1 million books - and uses the Apache Hadoop framework to facilitate preprocessing the large volume of data in order to create in-memory datasets used at runtime. Watson runs on an "embarrassingly parallel" architecture, says IBM's Dr. John Kelly, who led the effort. This "means that lots of different threads, different computations are triggered at the same time."
Beyond Jeopardy!, the deep Q&A technology behind Watson will be adapted to solve problems and drive progress in various fields, starting with healthcare, to help accurately diagnose patients. As Kelly put it: "We are trying to produce a deep question and answer machine that will change the way people interact with computers and machines. We're going to revolutionize many fields."
To stay on top of all the trends, subscribe to Database Trends and Applications magazine.