Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MIDDAG: Where Does Our News Go? Investigating Information Diffusion via Community-Level Information Pathways

Mingyu Derek Ma, Alexander K. Taylor, Nuan Wen, Yanchen Liu, Po-Nien Kung, Wenna Qin, Shicheng Wen, Azure Zhou, Diyi Yang, Xuezhe Ma, Nanyun Peng, Wei Wang

TL;DR

MIDDAG tackles the challenge of visualizing and forecasting information diffusion of COVID-19 news across social platforms by integrating user- and community-level analyses on Twitter and Reddit. The approach combines four components: community construction via the Influence-Passivity framework, IP forecasting with a graph neural network, susceptibility modeling using RoBERTa-based embeddings and a contrastive learning objective, and event extraction plus representative opinion mining with a T5-large, instruction-tuned model. Key contributions include a scalable community-based IP construction method with convergent $I_i$ and $P_i$ scores, an IP predictor achieving an AUC of $86.83\%$, a susceptibility model yielding $F_1 = 86.28\%$, and an end-to-end pipeline for extracting event triggers and dominant opinions. The work uses a cross-platform dataset (Twitter May 2020 subset and Reddit discussions) to deliver interactive visualizations and forecasting capable of informing researchers and practitioners about diffusion mechanisms, audience receptivity, and evolving crowd opinions.

Abstract

We present MIDDAG, an intuitive, interactive system that visualizes the information propagation paths on social media triggered by COVID-19-related news articles accompanied by comprehensive insights, including user/community susceptibility level, as well as events and popular opinions raised by the crowd while propagating the information. Besides discovering information flow patterns among users, we construct communities among users and develop the propagation forecasting capability, enabling tracing and understanding of how information is disseminated at a higher level.

MIDDAG: Where Does Our News Go? Investigating Information Diffusion via Community-Level Information Pathways

TL;DR

MIDDAG tackles the challenge of visualizing and forecasting information diffusion of COVID-19 news across social platforms by integrating user- and community-level analyses on Twitter and Reddit. The approach combines four components: community construction via the Influence-Passivity framework, IP forecasting with a graph neural network, susceptibility modeling using RoBERTa-based embeddings and a contrastive learning objective, and event extraction plus representative opinion mining with a T5-large, instruction-tuned model. Key contributions include a scalable community-based IP construction method with convergent and scores, an IP predictor achieving an AUC of , a susceptibility model yielding , and an end-to-end pipeline for extracting event triggers and dominant opinions. The work uses a cross-platform dataset (Twitter May 2020 subset and Reddit discussions) to deliver interactive visualizations and forecasting capable of informing researchers and practitioners about diffusion mechanisms, audience receptivity, and evolving crowd opinions.

Abstract

We present MIDDAG, an intuitive, interactive system that visualizes the information propagation paths on social media triggered by COVID-19-related news articles accompanied by comprehensive insights, including user/community susceptibility level, as well as events and popular opinions raised by the crowd while propagating the information. Besides discovering information flow patterns among users, we construct communities among users and develop the propagation forecasting capability, enabling tracing and understanding of how information is disseminated at a higher level.
Paper Structure (7 sections, 2 equations, 1 figure)

This paper contains 7 sections, 2 equations, 1 figure.

Table of Contents

  1. Introduction
  2. Data and System Design
  3. Components
  4. Community Construction
  5. Pathway Prediction
  6. Susceptibility
  7. Event Extraction and Community Opinion

Figures (1)

  • Figure 1: Visualizations of the user-level information pathways (left), community-level information pathways with estimated susceptibility level for each community (center), and community-level information propagation forecasting results (right).