Home // International Journal On Advances in Internet Technology, volume 14, numbers 1 and 2, 2021 // View article

A Topic Modeling Framework to Identify Online Social Media Deviance Patterns

Authors:
Thomas Marcoux
Esther Mead
Nitin Agarwal

Keywords: misinformation; disinformation; topic models; topic streams; COVID-19; misinfodemic; narratives

Abstract:
Following the COVID-19 pandemic and the subsequent vaccine related news, the information community has seen the emergence of unique misinformation narratives in a wide array of different online outlets, through social media, blogs, videos, etc. Taking inspiration from previous COVID-19 and misinformation detection related works, we expanded our topic modeling tool. We added filtering capabilities to the tool to adapt to more chaotic social media datasets and create a chronological representation of online text content. We curated a corpus of 543 misinformation pieces whittled down to 243 unique misinformation narratives, and collected two separate sets of 652,120 and 1,664,123 YouTube comments. From our corpus of misinformation stories, this tool has shown to accurately represent the ground truth of COVID misinformation stories. This highlights some of the misinformation narratives unique to the COVID-19 pandemic and provides a quick method to monitor and assess misinformation diffusion, enabling policy makers to identify themes to focus on for communication campaigns. To expand previous publications and further explore the potential of topic streams in understanding online misinformation, we propose a framework used as a filter to help whittle down big data corpora and identify latent misinformation within. This could be scaled and applied to very large social networks to highlight misinformation.

Pages: 60 to 72

Copyright: Copyright (c) to authors, 2021. Used with permission.

Publication date: December 31, 2021

Published in: journal

ISSN: 1942-2652