Allen Gorin on Stream Characterization
Allen Gorin: Stream Characterization from Content
Monday, April 06, 2009 | 4:00pm - 5:00pm
Deb Roy
Coping with information overload is a major challenge of the 21st century. Huge volumes and varieties of multilingual data must be processed to extract salient information. We have previously reported on research for how to automatically characterize large volumes of streaming content. However, information includes both content and associated meta-data, which humans deal with as a gestalt but computer systems often treat separately. Attributed random graphs provide a useful mechanism for jointly modeling content and context. This talk describes such methods, with experimental proof-of-concept on Switchboard and Enron corpora. This research is in collaboration with Priebe and Grothendieck.
Allen Gorin is Director of Human Language Technology (HLT) Research in the U.S. Department of Defense at Fort Meade, focused on creating HLT technologies for coping with information overload. Before that, he was at AT&T Labs, leading the research team that created AT&T's "How May I Help You?" natural language voice service, which was deployed nationally in 2001. He is a Fellow of the IEEE, was awarded the 2002 AT&T Science and Technology Medal, has published 101 papers, and been granted 33 U.S. Patents. He received the BS and MA degrees in mathematics from SUNY at Stony Brook, and the PhD in mathematics from the CUNY Graduate Center in 1980. He joined AT&T Bell Labs and led the DARPA ASPEN project, developing parallel architectures and algorithms for pattern recognition. In 1987, he was appointed Distinguished Member of the Technical Staff, then joined the Speech Research Department at Bell Labs in 1988. He joined the DoD in 2004. He was a visiting researcher at the ATR Lab in Japan in 1994 and at MIT in 2002.