Carterette receives prestigious national award for research on search engines
Search engines have come a long way since the first one, known as “Archie,” was launched almost 25 years ago. People no longer have to know the exact wording of a website’s title to find it, and searches now yield information not only on the specific topics requested but also on related subjects.
However, there’s still a long way to go in brokering the perfect marriage between what a user is looking for and what a search engine finds.
The University of Delaware’s Ben Carterette believes the key to making that marriage work is a better understanding of how the user and the system interact with each other, and he has been awarded a prestigious Faculty Early Career Development Award from the National Science Foundation (NSF) to support research aimed at solving that problem.
The $550,000 grant, “Measuring Search Engines’ Ability to Help Users Complete Tasks,” will support Carterette’s research and education program for the next five years.
“We use search engines for a wide range of tasks today, from planning a vacation to finding a good day care center,” says Carterette, associate professor in UD’s Department of Computer and Information Sciences. “Some of those tasks, like getting directions and or looking for a weather forecast, are relatively simple, while others, such as finding accurate medical information, are far more complex.”
“Also, users vary in their knowledge of how to use search engines and what to do with the information turned up in a search,” he adds. “Some people just automatically go right to the top item on the list, while others will sort through two or three pages of results to get information.”
Carterette plans to simulate the process a user goes through in conducting a search in an effort to improve the ability of the system to complete the desired task.
“Typically, users query the system, quickly read through the results to determine whether they see anything they like, and then initiate a new search or switch to a different search engine if the first search doesn’t yield anything promising,” he says. “Eventually, they’ll give up if they’re not successful in finding what they want.”
Carterette is hoping that his research will enable the number of steps in the process to be decreased, with users pointed more quickly in the right direction — or at least informed early in the process that they’re not going to be successful.
His proposed simulation will model user behavior so that search engines can be trained to better meet their needs.
“It’s basically an artificial intelligence question,” he says. “We know we can’t achieve a perfect simulation of human behavior, but we hope we can get close enough so that we can learn what we need to know to design better search engines.”
“Having a better understanding of what makes a search system useful will also lead to concrete improvements in search and human language technologies in general,” Carterette adds.
In addition to the research, the project will include an educational component, with a focus on increasing understanding of experimental design and statistical analysis at all levels of computer science education.
The highly competitive NSF Faculty Early Career Development Awards recognize junior faculty for their role as teacher-scholars and are given to those scientists and engineers considered most likely to become the academic leaders of the 21st century.
Article by Diane Kukich
Photo by Evan Krape