Design of machine learning powered visualizations to support rapid assessment of online student discussions

Priya Sharma, Qiyuan Li, Mahir Akgun

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper reports on the design of machine learning powered analytic visualizations to support instructors in rapidly assessing student behavioral and cognitive engagement in online discussions. We used Chi's (2009) Interactive-Constructive-Active-Passive framework to assess cognitive engagement and social network analysis to assess behavioral engagement. We used the Long-Short-Term-Memory (LSTM) model to automatically classify discourse data. The model was trained based on 4000 human-rater coded posts and despite imbalanced data, the model shows relatively high accuracy. Three use case scenarios of the visualizations show that network and discourse analyses together support the instructor in ascertaining students' cognitive and behavioral engagement. Next steps for addressing model accuracy and improving visualizations are presented.

Original languageEnglish (US)
Title of host publicationInternational Collaboration toward Educational Innovation for All
Subtitle of host publicationOverarching Research, Development, and Practices - 15th International Conference on Computer-Supported Collaborative Learning, CSCL 2022
EditorsArmin Weinberger, Wenli Chen, Davinia Hernandez-Leo, Bodong Chen
PublisherInternational Society of the Learning Sciences (ISLS)
Pages455-458
Number of pages4
ISBN (Electronic)9781737330646
StatePublished - 2022
Event15th International Conference on Computer-Supported Collaborative Learning, CSCL 2022 - Virtual, Online, Japan
Duration: Jun 6 2022Jun 10 2022

Publication series

NameProceedings of International Conference of the Learning Sciences, ICLS
ISSN (Print)1814-9316

Conference

Conference15th International Conference on Computer-Supported Collaborative Learning, CSCL 2022
Country/TerritoryJapan
CityVirtual, Online
Period6/6/226/10/22

All Science Journal Classification (ASJC) codes

  • Computer Science (miscellaneous)
  • Education

Fingerprint

Dive into the research topics of 'Design of machine learning powered visualizations to support rapid assessment of online student discussions'. Together they form a unique fingerprint.

Cite this