Intelligent Tutoring Systems Go Ill-Structured: Argument Instruction with Graphical Representations
Speaker
Vincent Aleven
Research Scientist, Human-Computer Interaction Institute, Carnegie Mellon University
When
-
Where
Newell-Simon Hall 1305 (Michael Mauldin Auditorium)
Description
Intelligent tutoring systems have been highly successful in “well-structured” domains such as math and science learning, but do not have a similar track record in domains or tasks that are “ill-structured.” In such domains, provably correct answers do not exist, and argumentation is often a primary mode of reasoning. Reasonable conflicting arguments occur, and educators evaluate their persuasiveness without necessarily resolving the conflict.
We report on some progress and early success in creating an intelligent tutoring system for an ill-structured domain, legal decision-making, as illustrated in oral arguments before the US Supreme Court. A key feature of these arguments is the posing of hypothetical situations to explore the ramifications of proposed rules for deciding a case.
We created an intelligent tutoring system, LARGO, that supports students in studying transcripts of US Supreme Court oral arguments. It supports students in creating diagrammatic representations of these types of arguments and is capable of identifying weaknesses and opportunities for reflection. In a study with 28 first-semester law students, we found that the use of LARGO did not lead to better learning of hypothetical reasoning skills across the whole sample, compared to a text-based note-taking system. LARGO however was more effective in helping lower-ability students learn according a number of key measures.
The LARGO project is joint work with Niels Pinkwart of Clausthal University of Technology, and Kevin Ashley and Collin Lynch of the University of Pittsburgh.
Speaker's Bio
Dr. Vincent Aleven is a Research Scientist in the Human-Computer Interaction Institute. He is most excited by research projects that both tell us something new about how people learn and provide a better way of supporting learners. His research focuses on intelligent tutoring systems, broadly defined as software “tutors” that use artificial intelligence to adapt instruction to the needs of individual learners or of collaborating groups of learners. The more specific questions driving his research are: (a) How can intelligent tutoring systems help students learn in ill-defined domains? (b) How can intelligent tutoring systems support students’ meta-cognitive processes, and with what effect on their domain-specific learning? (c) What kinds of authoring tools make intelligent tutoring systems easier to create?
Speaker's Website
http://www.cs.cmu.edu/~aleven/