Professor Ingrid Russell (far right) works with Marlon Gregory ´05 (left) and Malin Engman ´05 on a project in Russell´s fall AI class.

Ingrid Russell, professor of computer science in the College of Arts and Sciences, in collaboration with Zdravko Markov of Central Connecticut State University and Todd Neller of Gettysburg College, has received a $99,469 grant from the National Science Foundation's Course, Curriculum, and Laboratory Improvement (CCLI) program. Russell, who is the lead principal investigator on the grant, reports that it was one of only 10 funded this year from the approximately 90 computer science proposals submitted.

The team of three professors is using the grant to fund a machine-learning project that they hope will take the teaching of artificial intelligence (AI) into the 21st century. AI is the science and engineering involved in creating intelligent machines, especially computer programs.

"We are delighted that our project was selected for funding by a highly competitive program at NSF," says Russell. "We believe that this project will impact the way the traditional AI course is taught at many colleges. The project introduces students to an increasingly important research area in computer science and provides an opportunity for them to apply AI problem-solving techniques to a real-world application."

Russell is redefining an introductory one-semester AI course using machine learning to tie together diverse topics while developing a suite of adaptable, hands-on laboratory projects. Machine learning concerns developing computer systems or programs that can improve their performance based on previous experiences. It is increasingly used in science, engineering, information systems, and education for applications such as speech recognition, natural language processing, robotics, game playing, and medical data analysis.

Students work on the projects in teams to develop machine learning systems. In one project, Web User Profiling, students develop an intelligent Web browser, one that learns user preferences, to improve the efficiency of Web searches.

"Students use data mining [the process of extracting patterns from the data] and machine learning techniques to analyze samples of user interests or preferences in a given domain, such as movies/music, and create a profile of the user's interests," says Russell.

User profiling is used extensively in marketing to find patterns that can predict user purchases. Many retail stores collect data at the checkout scanner that is then used to determine everything from store inventories to what goods are displayed in the same or neighboring store aisles.

The success of this project has gone beyond the University of Hartford classroom. Three students on the Web User Profiling project-Shona Taiwo '05, Roberto Scata '07, and Richard Truncali '07-have had a paper accepted for presentation at the 10th Annual Consortium for Computing Sciences in Colleges Northeast Conference to be held in April.

Scata, a computer science major, recently received a NASA Undergraduate Fellowship to work with Russell on a research project that will extend the work done on the Web User Profiling project.

The course, which completed its first semester in the fall, received positive feedback from students. One student said, "Working on the project was a great experience. I was amazed by the wide range of applications of machine learning in various aspects of our lives."

Russell reports that several faculty members from computer science departments at other colleges and universities across the nation have already become affiliated with the project and committed to using the material being developed.