Using topic modeling to explore online discourse at a Twitter conference backchanne

• In Event: AECT Gallery Walk - D&D Showcase, Emerging Technology Showcase, AECT Gallery Walk - Available 24/7 during the convention.
  In Poster Session: GSA Poster Sessions

Short Description

Traditional ways to analyze large data can be time consuming and require extensive knowledge of the context and the community. We believe topic modeling provides a robust and swift way to explore contextual meaning based on keywords. In this study, we managed to reduce the number of tweet data for manual coding by 50%. However, any data mining result should not be the final stop. The “unbiased” results may not be meaningful to the decision makers. Additional qualitative inquiries on the representative data are needed to interpret the meanings as well as the validation to the data mining results.

Authors

• Presenter: HAJEEN CHOI, Florida State University

• Presenter: Zhichun Liu, Florida State University

• Presenter: Jiyae Bong, Concordia University, Montreal