Detecting student carefulness in an educational game for physics

Date of Award

2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Computer Science

Department

Information Systems & Computer Science

First Advisor

Rodrigo, Ma. Mercedes T., Ph.D.

Abstract

Carefulness is a construct that has been researched in the fields of education and social science. It is deemed as an important facet in learning as the more careful a student is, the less likely he/she will commit trivial errors or careless mistakes. Careful students have been seen to possess discipline more than students who are least careful. The general goal of this study is to create a detector for student carefulness in an educational game for Physics. A quantitative model for carefulness within Physics Playground is built and validated using semi-supervised learning, specifically self-training, to use both labeled and unlabeled data in building the carefulness detector. Nave bayes classification has been used as the modeling algorithm. Comparing the results of the iterations, it has been found that the models performance did not degrade, converged and resulted to improved predictions as compared to the base model/learner, which used purely labeled data or purely supervised nave bayes classification. Student samples from different cities in the Philippines have been used for this study. With self-training, carefulness was found to exist in the datasets of Philippine student samples and can be robustly detected by the features as hypothesized/modeled by Physics Playground developers and candidate features from related Social Science constructs as evidenced by the findings of this study.

Comments

The C7.B365 2018

Link to Full Text

Share

COinS