Automatically Understanding Handwritten Self-explanations

James Herold; Thomas Stahovich

Download Paper | Permalink

Conference: 2012 ASEE Annual Conference & Exposition
Location: San Antonio, Texas
Publication Date: June 10, 2012
Start Date: June 10, 2012
End Date: June 13, 2012
ISSN: 2153-5965
Conference Session: CoED General Technical Session II
Tagged Division: Computers in Education
Page Count: 10
Page Numbers: 25.246.1 - 25.246.10
DOI: 10.18260/1-2--21006
Permanent URL: https://peer.asee.org/21006
Download Count: 421

Paper Authors

biography

James Herold University of California, Riverside

visit author page

James Herold earned his B.S. in computer science at California Polytechnic State University, Pomona in 2004. He is currently a Ph.D. student in computer science at the University of California, Riverside.

visit author page

biography

Thomas Stahovich University of California, Riverside

visit author page

Thomas Stahovich received a B.S. in mechanical engineering from the University of California, Berkeley in 1988. He received a S.M. and Ph.D. in mechanical engineering from the Massachusetts Institute of Technology in 1990 and 1995, respectively. He is currently Chair and Professor in the Mechanical Engineering Department at the University of California, Riverside.

visit author page

Download Paper | Permalink

Abstract

Automatically understanding student self-explanations Research has demonstrated that written self-explanation can lead students to rethink their work,potentially increasing their understanding of the material. We have found, however, that not all self-explanation is substantive. Our goal is to develop computational techniques capable of determining if astudent’s explanation is relevant or not. This will then enable us to create an interactive system capableof prompting students to continue their explanations when necessary. This is a tractable task as thereare typically a small number of concepts expected for each explanation. The language used to expressthose concepts can vary greatly, but our task is only to identify the existence of the concepts, not toperform general machine interpretation. In this paper, we present early work on automaticallyunderstanding students' hand-written self-explanations describing their work on homework problems. In winter of 2011, we conducted a study in which 39 students in an undergraduate Statics coursewere asked to provide hand-written self-explanations of their work. Students were provided a set ofquestions with each homework assignment, eliciting explanation of their reasoning for each step of thesolution process. For example, students were asked why they chose the free body diagram they usedand, for friction problems, whether or not they assumed any bodies were slipping. Students wererequired to write homework solutions and self-explanations using an Anoto digital pen, creating a time-stamped digital copy of their work. In this paper, we employ information extraction techniques to automatically extract essentialrelations from the students’ self-explanations, enabling the machine to determine which concepts arepresent. For example, given a student’s self-explanation transcript, our software can determine if astudent assumed any bodies were on the verge of slip, and if so, which bodies. In our experiments, our technique has proven to be quite accurate. For example, in one case itwas more than 90% accurate at determining if students assumed bodies were on the verge of slip. Thislevel of accuracy is quite encouraging and paves the way for future work on intelligent tutoringsystems.

Citation
Format

Herold, J., & Stahovich, T. (2012, June), Automatically Understanding Handwritten Self-explanations Paper presented at 2012 ASEE Annual Conference & Exposition, San Antonio, Texas. 10.18260/1-2--21006

TY - CPAPER
AB - Automatically understanding student self-explanations Research has demonstrated that written self-explanation can lead students to rethink their work,potentially increasing their understanding of the material. We have found, however, that not all self-explanation is substantive. Our goal is to develop computational techniques capable of determining if astudent’s explanation is relevant or not. This will then enable us to create an interactive system capableof prompting students to continue their explanations when necessary. This is a tractable task as thereare typically a small number of concepts expected for each explanation. The language used to expressthose concepts can vary greatly, but our task is only to identify the existence of the concepts, not toperform general machine interpretation. In this paper, we present early work on automaticallyunderstanding students&#39; hand-written self-explanations describing their work on homework problems. In winter of 2011, we conducted a study in which 39 students in an undergraduate Statics coursewere asked to provide hand-written self-explanations of their work. Students were provided a set ofquestions with each homework assignment, eliciting explanation of their reasoning for each step of thesolution process. For example, students were asked why they chose the free body diagram they usedand, for friction problems, whether or not they assumed any bodies were slipping. Students wererequired to write homework solutions and self-explanations using an Anoto digital pen, creating a time-stamped digital copy of their work. In this paper, we employ information extraction techniques to automatically extract essentialrelations from the students’ self-explanations, enabling the machine to determine which concepts arepresent. For example, given a student’s self-explanation transcript, our software can determine if astudent assumed any bodies were on the verge of slip, and if so, which bodies. In our experiments, our technique has proven to be quite accurate. For example, in one case itwas more than 90% accurate at determining if students assumed bodies were on the verge of slip. Thislevel of accuracy is quite encouraging and paves the way for future work on intelligent tutoringsystems.
AU - James Herold
AU - Thomas Stahovich
CY - San Antonio, Texas
DA - 2012/06/10
PB - ASEE Conferences
TI - Automatically Understanding Handwritten Self-explanations
UR - https://peer.asee.org/21006
DO - 10.18260/1-2--21006
ER -