An information scientist gets personal
The small, Japanese-made eletric water heater in Peter Brusilovsky’s office is nothing like the elaborate, pear-shaped samovars of his native Russia. A centerpiece in Russian homes and a symbol of hospitality for more than 200 years, the towering samovars do more than heat tea. Their hearth-like character nurtures long conversation and keeps the tea flowing through a give-and-take of ideas, opinions, debate, and revolution.
Even so, the plastic pot in a corner of Brusilovsky’s office in Pitt’s School of Information Sciences gets the job done. The water is always hot and ready for a personal offering of tea as a backdrop and conduit for the free flow of information, thought, and counterthought. For at the core of Brusilovsky’s thinking and research are personalized information itself and the means by which people access it.
Information has been in oversupply since the advent of the World Wide Web. Search engines, metasearch engines, and directories—though powerful tools—are imprecise, with low accuracy rates. At fault is the inherent inaccuracy of the basic keyword linguistic processing used by information retrieval systems. Two or three keywords yield a fast response, but there is no interactive “listening” and “learning” in the information transaction. In a one-size-fits-all system, search engines don’t understand precisely what the user needs to know. At least not yet.
According to Brusilovsky, the issue is not merely access to information but personalized access. A leading innovator in the fields of adaptive hypermedia and intelligent tutoring systems, Brusilovsky, an associate professor of information science and telecommunications, is conducting research and developing curricula through a prestigious five-year National Science Foundation Faculty Early Career Development program grant. The award supports the development activities of teacher-scholars who effectively integrate research and education. Brusilovsky’s goal is to develop an integrated system for personalized access to education resources and content.
Adapting information to the needs of a computer user has been a challenge since the early 1980s. To increase the relevance of retrieved information, researchers have developed techniques of “user modeling” and “adaptive personalization.” Users rank and select information by relevance, providing feedback for retrieval systems to filter the flow of information.
Anyone who has shopped on Amazon.com has witnessed user profiling in action. Tastes and preferences regarding books, CDs, and DVDs are recorded, filtered, and translated into buying suggestions. People who have just placed a Star Wars prequel into their digital shopping carts are immediately alerted that others who bought the Star Wars DVD also purchased, for instance, Harry Potter and Lord of the Rings. Someone who is eager to buy the entire first season of Leave It to Beaver can receive advance notice by e-mail when it becomes available on DVD.
This type of low-level interactive filtering and retrieval based on user interest has limited effectiveness, however, in educational contexts. In learning situations, not only are the user’s interests required, but also the user’s specific goal, skill level, knowledge, and learning style. To help people access the vast sea of open information on the Web effectively, Brusilovsky is building on and expanding techniques that he and others have developed using adaptive navigation support and visualization technologies.
Students in Brusilovsky’s undergraduate computer programming courses already use “knowledge maps,” two-dimensional, computer-based diagrams that visually represent structure and content of subject matter. The prototype Knowledge Sea software groups similar documents about C programming language into map cells that students explore according to their own skill and knowledge levels.
Sitting at his computer after lunch, a cup of tea at hand, Brusilovsky demonstrates a practical application of Knowledge Sea. Guided by visual icons indicating which information is relevant, essential, confusing, or worth reading again, he navigates pages of text and examples. His browsing and marking activities update the program’s evolving profile of his knowledge. Brusilovsky envisions an expansion of this sea of C knowledge. In that extended sea, the course instructor will post lectures, notes, reference materials, and links to tutorials, digital libraries, and other relevant sites to a course Web site. Using sample problems and self-assessment quizzes, the program will extract concepts and index them on a personalized basis, progressively taking the student’s measure.
The new Knowledge Sea would serve as a portal to a widening hyperspace of information and examples. It represents the next generation of adaptive information retrieval and establishes a vital research program to study adaptive navigation support and visualization techniques applied to the Web.
Personalized access to information is gaining expressive power to order and compartmentalize the e-info glut. New information technologies may be a far cry from the communal interaction of lessons shared in the radiance of a copper-clad Russian samovar. But by building on the human tradition of learning through interaction, Peter Brusilovsky will continue to explore and map the knowledge sea.
Breakthroughs in the Making
Since 1900, when rotary drills replaced cable tools, there haven’t been any revolutionary changes in the technology to drill through rock—until now.
New joint research led by the Gas Technology Institute (GTI), with the U.S. Department of Energy and industry sponsors, suggests that advances in fiber laser technology hold promise for more efficient and less costly alternatives to gas and oil well construction techniques.
Managing the research at a new facility in Des Plaines, Ill., is petroleum and chemical engineer Brian Gahan (KGSB ’85). Using a 5-kilowatt fiber laser, the largest available for research in the United States, Gahan and his team found that high-power lasers quickly penetrate all rock, including granite, without the conventional use of drill bits and explosives. Tests also show that high-power lasers increase permeability and porosity of rock, improving fluid flow to the well. With yearly well-drilling depths totaling three times the diameter of the Earth, laser applications are likely to reduce costs in oil and gas exploration and production.
“For the future,” says Gahan, “I see applications in mining, quarrying, space exploration, and military operations. High-power fiber laser technology opens doors to applications far beyond the lab.”