Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

From Skepticism to Acceptance: Simulating the Attitude Dynamics Toward Fake News

Yuhan Liu, Xiuying Chen, Xiaoqing Zhang, Xing Gao, Ji Zhang, Rui Yan

TL;DR

This study introduces FPS, an LLM-based agent framework to simulate attitude dynamics toward fake news at micro and macro scales. Each agent possesses persona-based traits, dual memory, and a reflective reasoning process, enabling nuanced, text-rich simulations of opinion shifts. The framework couples a Dynamic Opinion Agent with an Agent Interaction Simulator to model daily interactions, topic dynamics, and interventions via an official spokesperson. Findings show topic and trait influence on propagation, and that early, appropriately frequent interventions effectively balance governance cost and effectiveness, offering practical guidance for misinformation management.

Abstract

In the digital era, the rapid propagation of fake news and rumors via social networks brings notable societal challenges and impacts public opinion regulation. Traditional fake news modeling typically forecasts the general popularity trends of different groups or numerically represents opinions shift. However, these methods often oversimplify real-world complexities and overlook the rich semantic information of news text. The advent of large language models (LLMs) provides the possibility of modeling subtle dynamics of opinion. Consequently, in this work, we introduce a Fake news Propagation Simulation framework (FPS) based on LLM, which studies the trends and control of fake news propagation in detail. Specifically, each agent in the simulation represents an individual with a distinct personality. They are equipped with both short-term and long-term memory, as well as a reflective mechanism to mimic human-like thinking. Every day, they engage in random opinion exchanges, reflect on their thinking, and update their opinions. Our simulation results uncover patterns in fake news propagation related to topic relevance, and individual traits, aligning with real-world observations. Additionally, we evaluate various intervention strategies and demonstrate that early and appropriately frequent interventions strike a balance between governance cost and effectiveness, offering valuable insights for practical applications. Our study underscores the significant utility and potential of LLMs in combating fake news.

From Skepticism to Acceptance: Simulating the Attitude Dynamics Toward Fake News

TL;DR

This study introduces FPS, an LLM-based agent framework to simulate attitude dynamics toward fake news at micro and macro scales. Each agent possesses persona-based traits, dual memory, and a reflective reasoning process, enabling nuanced, text-rich simulations of opinion shifts. The framework couples a Dynamic Opinion Agent with an Agent Interaction Simulator to model daily interactions, topic dynamics, and interventions via an official spokesperson. Findings show topic and trait influence on propagation, and that early, appropriately frequent interventions effectively balance governance cost and effectiveness, offering practical guidance for misinformation management.

Abstract

In the digital era, the rapid propagation of fake news and rumors via social networks brings notable societal challenges and impacts public opinion regulation. Traditional fake news modeling typically forecasts the general popularity trends of different groups or numerically represents opinions shift. However, these methods often oversimplify real-world complexities and overlook the rich semantic information of news text. The advent of large language models (LLMs) provides the possibility of modeling subtle dynamics of opinion. Consequently, in this work, we introduce a Fake news Propagation Simulation framework (FPS) based on LLM, which studies the trends and control of fake news propagation in detail. Specifically, each agent in the simulation represents an individual with a distinct personality. They are equipped with both short-term and long-term memory, as well as a reflective mechanism to mimic human-like thinking. Every day, they engage in random opinion exchanges, reflect on their thinking, and update their opinions. Our simulation results uncover patterns in fake news propagation related to topic relevance, and individual traits, aligning with real-world observations. Additionally, we evaluate various intervention strategies and demonstrate that early and appropriately frequent interventions strike a balance between governance cost and effectiveness, offering valuable insights for practical applications. Our study underscores the significant utility and potential of LLMs in combating fake news.
Paper Structure (16 sections, 2 equations, 5 figures, 1 table, 1 algorithm)

This paper contains 16 sections, 2 equations, 5 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Method
  4. Problem Formulation
  5. Our Simulation Framework
  6. Dynamic Opinion Agent
  7. Agent Interaction Simulator
  8. Experiments
  9. Implementation Details
  10. Metrics
  11. Macro-level Observation
  12. Micro-level Observation
  13. Analysis and Discussion
  14. Fake News Intervention
  15. Ablation Study
  16. ...and 1 more sections

Figures (5)

  • Figure 1: (a) Previous macro-level fake news modeling focused on predicting the overall infected population but lacked a detailed analysis of the dynamics in human attitudes. (b) Previous micro-level models translated human opinions and communication into numerical values, i.e., $\theta \in \mathbb{R}$ and $x \in \mathbb{R}$. (c) Our micro-level simulation uniquely captures attitude changes through natural language processing. (d) Additionally, our multiple agents constitute a macro-level simulation that also enables the prediction of popularity trends.
  • Figure 2: Our framework equips each agent with reasoning and response capabilities by creating a feedback loop between dynamic opinion agents (DOA) and an agent interaction simulator (AIS). 'Susceptible' denotes individuals skeptical of the fake news, 'infected' refers to those who believe in the fake news, and 'recovered' means individuals who previously believed the fake news but now do not.
  • Figure 3: Dynamic group population number changes in terms of different topics and traits, with an accompanying fitting curve based on the SIS model. The red dashed line represents the results of the SIS model fitting, where $\beta$ is the transmission rate and $\gamma$ is the recovery rate.
  • Figure 4: Comparison of models with different office agent intervention strategies. It can be seen that early and reasonably frequent regulation on fake news can lead to a significant reduction in its propagation and influence.
  • Figure 5: Micro-level case study of two people of different traits. Michael, possessing a credulous nature, frequently changes his opinions, whereas Sandra, being skeptical, tends to maintain a consistent view.