Date of Award

8-31-2020

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Information Systems - (Ph.D.)

Department

Informatics

First Advisor

Yi-Fang Brook Wu

Second Advisor

Starr Roxanne Hiltz

Third Advisor

Michael Bieber

Fourth Advisor

Shaohua David Wang

Fifth Advisor

Bodong Chen

Abstract

As a pedagogical strategy, Writing-to-Learn (WTL) intends to use writing to improve students’ understanding of course content. However, most of the existing feedback systems for writing are mainly focused on improving students’ writing skills rather than their conceptual development. In this dissertation, an automatic approach is proposed to generate timely, actionable, and individualized feedback based on comparing knowledge representations extracted from lecture slides and individual students’ writing assignments. The novelty of the proposed approach lies in the feedback generation: to help students assimilate new knowledge into their existing knowledge better, their current knowledge is modeled as a set of matching concepts; suggested concepts and concept relationships for inclusion are generated as feedback by combining two factors, i.e., importance and relevance, of feedback candidates to the matching concepts in the domain knowledge. In the prototype system, a student can request feedback many times; each set of feedback is generated for a corresponding assignment draft to support their learning of conceptual knowledge during the iterative process of writing an assignment.

This research conducts a repeated measures study across two semesters (N=88) to understand how students perceive the proposed system, explore how students use the automated feedback, and investigate the effects of the automated feedback on student learning. Survey results show that the feedback is perceived as relevant (78.4%), easy to understand (82.9%), accurate (76.1%) and useful (79.5%); survey results also find that the proposed system makes it easier to study course concepts (80.7%) and is useful in learning course concepts (77.3%). Based on the log analysis of students’ actual usage of the system, all participants request feedback at least once when using the proposed system. After requesting feedback, 83 out of 88 participants revise their assignments. Analyses of students’ submitted assignments reveal that more course concepts and concept relationships are included when completed using the proposed system. Collectively, these results show that the proposed automated feedback prototype system contributes to students incorporating more course concepts and concept relationships into their writing assignments, thus supports their learning of conceptual knowledge in a WTL activity.

Share

COinS