Learning About Learning: Using Intelligent Tutoring Systems to Study Human Learning

Abstract

Intelligent Tutoring Systems have been shown to help students learn more. For instance, the Carnegie Learning Tutoring system is used in 4% of American High Schools and has been shown to lead to large learning gains. Despite their success, there are very few of these systems in existence. The Office of Naval Research funded us to build authoring tools that should make it easier to create intelligent tutoring system. We even want regular teachers to be able to create their own content. In this talk I will explain some of the interdisciplinary work done in computer science, data mining, cognitive science, psychometrics and web-based development. Then I will explain how we used the authoring tool for a project for the US Dept of Education that assess students' state test scores (i.e. a student's Massachusetts Comprehensive Assessment System score (MCAS)) while at the same time tutors students. I will describe some of the data mining research we did that is incorporated into the system. I will also describe our NSF funded efforts at building better fitting cognitive models that model ~100 8th grade math skills per student using Bayesian networks. I hope to demonstrate to young computer scientists that you can both help public schools while at the same time do some interesting rigorous research.

Bio Dr. Neil Heffernan graduated summa cum laude from Amherst College in History and Computer Science. Neil taught mathematics to eighth grade students in Baltimore City as part of Teach for America, a program that selectively recruits top candidates to teach in inner-city schools. Neil then decided to do something easier and get a PhD in building intelligent tutoring systems. Neil enrolled in Carnegie Mellon University's Computer Science Department to do multi-disciplinary research in cognitive science and computer science to create educational software that leads to higher student achievement. For his dissertation, Neil built the first intelligent tutoring system that incorporated a model of tutorial dialog. As a post-doc, Neil managed a team of programmers and PhDs to create authoring tools to make it easier to build intelligent tutoring systems. Neil was awarded a National Academy of Education Postdoctoral Research Fellow funded by the Spencer Foundation. Neil is now an assistant professor at Worcester Polytechnic Institute, where he is focused on creating "cognitive models", computer simulations of student thinking and learning, which are then used to design educational materials, practices and technologies. Neil is working in close collaboration with the Worcester Public Schools, teams of teachers and WPI graduate students to create the next generation of intelligent tutoring systems, which are currently being used by a few thousand 8th grade students in and around Worcester. Since coming to WPI, Neil has received 6 grants (i.e, NSF CAREER, US Dept of Education, Office of Naval Research, the US Army, and the Spencer Foundation) worth over 3 millions dollars.