August 2, 2010
“We used controlled experiments to develop a model for the way people use a computer mouse when they plan to make a purchase,” explains Agichtein, assistant professor of mathematics and computer science. “When we apply this shopping model to data from actual Web users in an uncontrolled environment, it correlates to a doubling of the ad click-through rate.”
Agichtein and Guo, a graduate student, will present their findings this month at the SIGIR 2010 conference in Geneva, Switzerland. SIGIR is the leading forum for innovations in information retrieval and Web search.
The patterns of computer mouse behavior vary a great deal, depending on the habits of the user, Agichtein says. “A lot of skeptics believed that mouse movement is too varied to be useful, but our study validates that it can be a valuable indicator of a searcher’s intent.”
Agichtein heads the Emory Intelligent Information Access Lab.