Date of Award

Spring 2014

Degree Type

Thesis

Degree Name

Master of Science (MS)

Department

Computer and Information Technology

First Advisor

Alejandra J. Magana

Committee Member 1

Bedrich Benes

Committee Member 2

Grant P. Richards

Abstract

Misconceptions are commonly encountered in many areas of science and engineering where a to-be-learned concept conflicts with prior knowledge. Conceptual change is an approach for identifying and repairing the misconceptions. One of the ways to promote student conceptual change is providing students with ontological schema training. However, assessment of conceptual change relies on qualitative analysis of student responses. With the exponential growth of qualitative data in the form of graphical representations or written responses, the process of data analysis relying on human experts has become time-consuming and costly. This study took the advantages of natural language processing and machine learning techniques to analyze the responses effectively. In addition, we identified how students described complex phenomena in thermal and transport science and compared the differences of descriptions between students who took certain training courses to address misconceptions by means of ontological schema training and those who were exposed to a different course about the nature of science. After comparing the effectiveness of three different text classification methods - query-based approach, Naive Bayes classifier, and support vector machine (SVM) for identifying conceptual change, SVM classifier was chosen to assess student responses from a corpus collected by Streveler and her research group in previous studies. Based on the automatic assessment for student conceptual change, this research found that training students with appropriate ontological schema would promote the conceptual change.

Share

COinS