Watson's Lead Developer: 'Deep Analysis, Speed, and Results'
Computing Community Consortium (06/07/11) Erwin Gianchandani
The challenge of creating a machine that could compete on Jeopardy! was the genesis of IBM's Watson supercomputer, said Watson lead developer David Ferrucci in a keynote speech at ACM's 2011 Federated Computing Research Conference. He said the challenge involved balancing computer programs that are natively explicit, fast, and exciting in their calculations with natural language that is implicit, innately contextual, and frequently inaccurate. Ferrucci also said that it offered a "compelling and notable way to drive and measure technology of automatic question answering along five key dimensions: Broad, open domain; complex language; high precision; accurate confidence; and high speed." An analysis of tens of thousands of randomly sampled Jeopardy! clues revealed that the most common clue type occurred only 3 percent of the time, and this finding led to several guiding principles for Watson's development, including the unsuitabilty of specific large, hand-crafted models, the need to derive intelligence from many diverse techniques, and the primary requirement of massive parallelism. Ferrucci's team eventually designed a system that produces and scores numerous hypotheses using thousands of natural language processing, information retrieval, machine learning, and reasoning algorithms in combination. The research expanded the field of computer science by pursuing goal-oriented system-level metrics and longer-term incentives.
Tuesday, June 7, 2011
Blog: Watson's Lead Developer: 'Deep Analysis, Speed, and Results'
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment