Inferring Organization of Chemistry Knowledge from Language
Research Mentor(s)
Hutchinson, Brian
Description
Education researchers are deeply interested in understanding the way students organize their knowledge, and the way this organization evolves as a student gains expertise. Card sort tasks, which require students to group concepts, are one mechanism to infer organizational strategy, but their limited resolution means they necessarily miss some of the nuance in a student’s strategy.In this work,we propose a transformer network-based approach that leverages a richer signal: natural language justifications of sorts. Evaluated using data from a university chemistry card sort task, we find our approach obtains 74.8% accuracy on a six-way classification task identifying the card group described, and 84.2% on a binary classification task determining whether a group belonged to a principle-or representation-based strategy, far exceeding the majority class baselines of 20.9% and 56.8%, respectively.
Document Type
Event
Start Date
18-5-2020 12:00 AM
End Date
22-5-2020 12:00 AM
Department
Secondary Education
Genre/Form
student projects, posters
Subjects – Topical (LCSH)
Knowledge and learning; Knowledge, Theory of; Knowledge management
Type
Image
Rights
Copying of this document in whole or in part is allowable only for scholarly purposes. It is understood, however, that any copying or publication of this document for commercial purposes, or for financial gain, shall not be allowed without the author’s written permission.
Language
English
Format
application/pdf
Inferring Organization of Chemistry Knowledge from Language
Education researchers are deeply interested in understanding the way students organize their knowledge, and the way this organization evolves as a student gains expertise. Card sort tasks, which require students to group concepts, are one mechanism to infer organizational strategy, but their limited resolution means they necessarily miss some of the nuance in a student’s strategy.In this work,we propose a transformer network-based approach that leverages a richer signal: natural language justifications of sorts. Evaluated using data from a university chemistry card sort task, we find our approach obtains 74.8% accuracy on a six-way classification task identifying the card group described, and 84.2% on a binary classification task determining whether a group belonged to a principle-or representation-based strategy, far exceeding the majority class baselines of 20.9% and 56.8%, respectively.