In this talk I will discuss my ongoing work on the automatic grading of student-produced arguments. Argumentation is a difficult skill for novices that cuts across disciplines and age groups. Arguments are also uniquely difficult for intelligent systems due to their complex, and often implicit structure, as well as their representation as text or spoken words. Interest in automatic grading has grown in recent years with the rise of national standards such as Common Core which emphasize argumentation, and an increasing demand for personalized learning and support systems.

I will describe a data-driven approach to the problem based upon representing arguments graphically using argument diagrams; evaluating those diagrams using Augmented Graph Grammars; and inducing new grammars via Evolutionary Computation. This approach has the advantage of scaffolding both the student that is making arguments, and the system that is analyzing them.

Collin F. Lynch is a Research Assistant Professor of Computer Science at North Carolina State University. He received his PhD in Intelligent Systems from the University of Pittsburgh. His research areas include: graph-based educational data mining, intelligent tutoring systems for ill-defined domains, data-driven tutoring, student modeling, and social networks in MOOCs and blended courses. Dr. Lynch was a charter member of the International Educational Data Mining Society and currently serves as its Policy Chair.