Towards learning outcome prediction via modeling question explanations and student responses

Abstract

Predicting the current learning outcome of a student based on his/her responses to previous questions is a vital task for personalized education. Deep learning based models have achieved satisfactory performance on this task as they can automatically extract meaningful signals related to student learning outcome predictions. However, these models are unable to leverage the following critical pieces of information. First, the correct choice to a question is often accompanied by a textual explanation on why this choice is correct. This explanation text contains valuable information that could help better model the students’ mastery of needed concepts towards correctly answering the question. Second, existing models only consider whether a student’s response is correct or not (i.e., learning outcome) but do not distinguish between the wrong choices. In fact, which wrong choice a student has made in his historical responses can still provide rich information regarding student’s future learning outcome. Motivated by these facts, we propose a novel deep learning framework named DSLOP which effectively integrates both question explanations and students’ responses for predicting students’ learning outcome. The framework consists of four modules, i.e., a text encoder, a question encoder, a student encoder and a learning outcome predictor, which interact with each other to simulate how students’ responses are made based on his capabilities and questions’ characteristics. Particularly, we integrate question explanations and students’ responses into the DSLOP framework by using them as constraints for supplementary supervision. To evaluate the effectiveness of the proposed framework, comprehensive experiments are performed on datasets collected from a learning platform for exam preparation. The superior performance of the proposed DSLOP framework demonstrates its effectiveness and the necessity of using the question explanations and students’ responses in student learning outcome prediction task.