123456                               james dabrowiak, robert burdickphoto
Professor Joseph Chaiken helped create the Photochromic Optical Memory System (a working prototype is pictured above), based on a patented data storage process that uses lasers.

      Like Chaiken, Elizabeth D. Liddy G'88 loves the balance between pedagogical and entrepreneurial pursuits. "They're mutually supportive," says Liddy, co-founder, president, and CEO of TextWise LLC, a five-year-old research-and-development firm based at the CASE Center that specializes in information-retrieval technology. "We're so much on the leading edge that it makes teaching, from my perspective, and learning, from the students' perspective, really exciting."
      The School of Information Studies professor has good reason to be excited. A former reference librarian, she turned her frustrations with online searches for information into a booming business venture based on a natural language processing (NLP) system. This system adds a human-like understanding to queries by employing several levels of linguistic analysis in retrieving text. While many Internet search engines merely match words and yield a flood of unrelated material, Liddy's NLP software system, known as DR-LINK, sifts through documents with an analytical touch that can do everything from evaluate commentaries to match euphemisms with their unpleasant counterparts.

Woojin Paik, who serve as the company's directors of engineering and research, respectively. TextWise, with about 30 employees, is a sister company to Manning & Napier Information Services, which handles product marketing, sales, and customer service for TextWise. Along with DR-LINK, the range of TextWise software in the works or on the market includes: MAPIT (searches and analyzes patent databases); CHESS (extracts requested information, builds an expanding knowledge base of it, and tracks it chronologically over time); CINDOR (retrieves information in different languages and includes a translation); EVA (acts as an "intelligent agent" system by learning a user's interests and sources and providing updated information, including maps and other visuals, on specific topics); and KNOW-IT (gathers and organizes information in a knowledge base with visual displays).
      On a recent day, Liddy had just returned from California, where she'd dazzled Department of Defense and military officials with demonstrations of TextWise's information-retrieval technology. A board on the wall read: "Congrats 98K grant from NIH." Yes, the National Institutes of Health had awarded TextWise a grant to create a DR-LINK medical application. Seemingly in constant motion, she motored between the company's administrative and research floors in the CASE Center, answering questions on the fly. When asked where all this technology will lead, she cites the information boom of e-mail, voice mail, faxes, and the Internet. "My hope is it will become more and more second nature, so you think about what you're doing instead of the technology. The technology should become so transparent that you're not aware of it," she says. "When you need information, it will be there. And the important thing is it's the right information. That's why we need natural language processing technology, because it can get you very precise results without all the unnecessary excesses."





1234

quote

123456 "The goal is to help users make sense of the information glut," she says. "Language is quite ambiguous; we take language at all of these levels at which humans extract meaning and have the computer do the same thing."
      Liddy currently has seven patents in the works connected to DR-LINK and related technologies. In fact, DR-LINK, which evolved out of Liddy's doctoral degree work at Syracuse University and earned her three national and international awards, includes among its users the U.S. Patent and Trademark Office, the government's intelligence community, and major corporations. Liddy launched TextWise to commercialize DR-LINK in 1993 with her research partners, computer and information science doctoral student Edmund Yu and IST doctoral student
needs_for_[neural_networks]
The importance of precision should never be ignored. Take aluminum production, for instance. Several years ago, Alcan Rolled Products Company in Oswego, New York, offered engineering professor Can Isik a challenge: It wanted to more accurately control the thickness of aluminum during the production process, which involves rolling large coils of aluminum to thin it out. The control system in use relied on actual measurements taken near the end of the process, so when variations occurred there was often a delay in correcting thickness. "By the time you'd find out there was something wrong, it was almost too late," Isik says.





1234


Continued on page 6
Continued on page 7
Back to page 1
Back to page 2
Back to page 3
Back to page 4



Main Home Page Winter 1998-99 Issue Contents
Chancellor's Message Opening Remarks In Basket
Pan Am 103 Architecture at 125 Inventive Minds
Multi-Majors Quad Angles Campaign News
Student Center Faculty Focus Research Report
Alumni News/Notes View From The Hill University Place


E-mail the magazine
E-mail the web guy
SYRACUSE UNIVERSITY MAGAZINE
820 Comstock Ave., Rm. 308
Syracuse, NY 13244-5040