Web 2.0 Conference 2005

Workshop

Data on the Move (Sponsored by AT&T)

Rich Cox, Member, AT&T Research Labs

Date: Wednesday, October 5th, 2005
Time: 1:30pm - 2:45pm
Location: Olympic

Additional panelists:
Mazin Gilbert, AT&T
Suranga Chandratillake, Blinkx
Usama Fayyad, Yahoo!
Dave Sherfesse, Alexa

The Internet is the world's largest source of information, offering data on nearly every topic imaginable. The challenge is to extract useful information from the constant and enormous daily flow--in essence, to do data mining on a moving target, "in flight" as it were.

Web data comes in formats ranging from traditional fielded records, through text, speech, image, and video. The bulk of the data is referred to as "semi-structured" because it inherits the hypertext structure of the Web; organizationally, however, more often than not it is unstructured text.

The magnitude and richness of this data offers a compelling target for data mining. One can derive information about nearly any subject or anyone. Companies involved in web-based commerce --virtually all businesses these days--present a tremendous volume of information about their products, services, and operations. More importantly, one also can learn a tremendous amount about individual web users.

The researchers at AT&T Labs are now turning their data mining expertise to the challenges presented on the Web. AT&T's Daytona Database Management System is the largest known database in the world, with more than 100 terabytes of data and one trillion indexed records. For years, Daytona has been used to spot telephone calling card fraud in real time, and it is now being used in AT&T's Internet Protect service to analyze the 2,500 terabytes of IP traffic that flows over the company's backbone network every day to spot early signs of worm and virus attacks.

AT&T will host a panel discussion with other industry experts on technologies and applications of data mining, presenting views from across the industry on the use of data available on the Web and data created by use of the Web.

Workshop

Media Sponsors