Human-Computer Interaction Thesis Defense

  • Remote Access Enabled - Zoom
  • Virtual Presentation
  • XU WANG
  • Ph.D. Student
  • Human-Computer Interaction Institute
  • Carnegie Mellon University
Thesis Orals

Sourcing Open-ended Solutions to Create Scalable Learning Opportunities

Inreasing numbers of people are seeking higher education and professional development. A challenge to meet the growing demand is to scale such educational opportunities while maintaining their quality. My work tackles this challenge by harnessing examples from existing repositories to enable the creation of scalable and quality educational experiences. Learning sciences theories suggest that active learning that encourages students to participate and engage with the content is better than passive lecturing, and that focused practice targeting specific skills, appropriate scaffolding with timely feedback helps novices become experts. However, in practice, as we may have observed in a higher education context, it is not easy for the design of learning experiences to carry out the above characteristics. I explore ways to leverage the complementary strengths of experts, novices and machines to support the creation of high quality learning opportunities at scale. 

In this dissertation, I present insights from the iterative design, development and testing process of a technique UpGrade, over the course of 3 years engaging with 10 instructors and 500+ students at CMU. UpGrade is a technique that supports the semi-automatic creation of multiple-choice questions leveraging past students’ written solutions. I present findings on the learning benefits of UpGrade-produced questions through three classroom experiments. I then present findings from a series of studies regarding the instructional choice of when is appropriate to use evaluation-type exercises, such as multiple-choice questions, for student learning. I will also describe the design decisions along the way to incorporate instructors’ preferences and current workflows in writing questions into the authoring interface of UpGrade.

My work makes contributions to Learning Technologies, Learning Sciences and Human-Computer Interaction. On the learning sciences and technologies side, I discuss insights around supporting instructors in making more nuanced and accurate instructional decisions, and building tools to support instructors author active learning activities. On the HCI side, I discuss insights on methods to collect and harness resources for scalable content creation, and suggest future pathways to better leverage the complementary strengths of peers, experts and machine intelligence to support learning at scale.

Thesis Committee:
Kenneth Koedinger (Co-Chair)
Carolyn Rose (Co-Chair, HCII/LTI)
Jeffrey Bigham
Chinmay Kulkarni
Scott Klemmer (University of California, San Diego)

Addtitional Thesis Information

Zoom Participation Enabled. See announcement.

For More Information, Please Contact: 
Keywords: