News

Ron Brachman Honored at AAAI-08


EC08

A big piece of news from the Twenty-Third AAAI Conference on Artificial Intelligence (AAAI-08), which took place July 13 to 17 in Chicago, was the Distinguished Service Award being presented to Ron Brachman, vice president of worldwide research operations for Yahoo! Research.

Brachman, who is a founding fellow of AAAI, the source of numerous significant technical contributions to the field, and co-author of a major textbook, was honored for his extraordinary service to the AI community. In particular, Brachman was recognized for “his contributions to the field of artificial intelligence through sustained service in numerous leadership roles in industry, government and professional societies, tirelessly instigating, facilitating and promoting successful AI research and development.”

Brachman’s contributions to the field are legendary: he was recently President of AAAI; was Secretary-Treasurer of IJCAI for a decade and Director of the Information Processing Technology Office at DARPA (where he funded a billion dollars’ worth of Computer Science research); was co-founder of the KR series of conferences now in its twentieth year; and is the inventor of Description Logics, a key element in the emerging Semantic Web. The significant AAAI award follows on the heels of Brachman’s receiving IJCAI’s Donald E. Walker Distinguished Service Award in 2007.

The annual AAAI conference is one of the largest and most prestigious conferences in artificial intelligence. The event promotes research in AI and scientific exchange among researchers, practitioners, scientists and engineers.

A bronze sponsor of this year’s event, Yahoo! Research presented numerous papers and workshops covering a wide area of artificial intelligence.

Evgeniy Gabrilovich co-organized a day-long workshop called, “Wikipedia and Artificial Intelligence: An Evolving Synergy.” The popular workshop highlighted the mutually beneficial interaction between Wikipedia and artificial intelligence, and showed how Wikipedia can be a valuable resource for a diverse set of AI applications.

The workshop featured an invited talk by Michael Witbrock of Cycorp, the company behind the well-known CYC project. Witbrock noted that with today’s volume of information, it is not always required – or even possible – to employ humans to verify the accuracy of Wikipedia submissions. Instead, he said, it’s preferable to use computers in simulated environments to verify facts by repeatedly observing the behaviors of others, and assuming repeated behaviors to be correct.

Yahoo! also took part in the AAAI Intelligent Systems Demonstrations program, which showcased state-of-the-art AI implementations and provided researchers with an opportunity to demonstrate their work. Selected for the program was a project created by Yahoo! researchers Sharad Goel, David Pennock, Daniel Reeves, and Cong Yu called “Yoopick: A Combinatorial Sports Prediction Market.”

Yoopick implements a flexible language that allows sports aficionados to pick the final point spread of any game. For example, a player might pick a team to win by between 2 and 11 points. The more a prediction is viewed as unlikely by others, and the more the fan is willing to stake on his prediction, the more he stands to gain. Yoopick attracted a steady stream of visitors interested in both the science behind the application as well as the slick interface.

Yahoo! also presented papers at AAAI-08. Accepted papers either written or co-written by Yahoo! researchers included

  • An Expressive Auction Design for Online Display Advertising
    Sebastien Lahaie, David Parkes, and David Pennock
  • Concept-Based Feature Generation and Selection for Information Retrieval
    Ofer Egozi, Evgeniy Gabrilovich (Yahoo!), and Shaul Markovitch

There were a number of interesting talks presented by others as well: one showed how to design a voting rule that discourages participants from voting more than once, which is good for the democratic process, but bad for American Idol. David Haussler described how computational methods can now be used to infer the DNA of our common ancestors from thousands of mammals, shedding new light on how the human brain evolved. Tom Mitchell summarized his recent work, published in Science, in which, using machine learning, he analyzed images of brain activity, and demonstrated that the same word activates roughly the same brain regions in different people and that the same concept described in different languages triggers similar patterns of brain activity.

Conference attendees also marveled at the robot exhibit, which was one of the most visually striking. It included robot snakes, robot dogs, robot spiders, re-purposed robot vacuums, and yellow bouncy snowmen that dance in tandem. Meanwhile, in the ever-popular poker competition, computer players continued to improve their head-to-head play, now approaching human expert level, and for the first time ever competed well in six-player games.

Another conference highlight was the annual AI Video Competition, which was jointly sponsored by Yahoo! Research, Microsoft Research, and Google. The competition, also known as the Oscars of AI research videos, documents some of the most exciting artificial intelligence advances in a humorous and creative fashion.

This year’s conference clearly demonstrated how much fun AI is, and how AI research is changing the world.