Winning the Social Media Influence Battle: Uncertainty-Aware Opinions to Understand and Spread True Information via Competitive Influence Maximization

TL;DR

The paper addresses competitive influence maximization in online social networks under uncertainty by introducing a dual-DRL framework guided by Subjective Logic. It replaces binary opinions with a Uncertainty-aware Opinion Model (UOM) and compares three opinion models, examining how user trust and engagement affect seed-node selection and information quality. The framework employs PPO-based training with six strategies for the false party and evaluates on the URV Email Network, showing that UOM consistently improves true-information spread by over 20% and remains robust under partial observability; DRIM-A and DRIM-NA offer strong efficiency-performance tradeoffs. The work advances CIM by integrating nuanced opinion dynamics and partial observability, with practical implications for mitigating misinformation in real-world OSNs and guiding scalable diffusion strategies.

Abstract

Competitive Influence Maximization (CIM) involves entities competing to maximize influence in online social networks (OSNs). Current Deep Reinforcement Learning (DRL) methods in CIM rely on simplistic binary opinion models (i.e., an opinion is represented by either 0 or 1) and often overlook the complexity of users' behavioral characteristics and their prior knowledge. We propose a novel DRL-based framework that enhances CIM analysis by integrating Subjective Logic (SL) to accommodate uncertain opinions, users' behaviors, and their preferences. This approach targets the mitigation of false information by effectively propagating true information. By modeling two competitive agents, one spreading true information and the other spreading false information, we capture the strategic interplay essential to CIM. Our framework utilizes an uncertainty-based opinion model (UOM) to assess the impact on information quality in OSNs, emphasizing the importance of user behavior alongside network topology in selecting influential seed nodes. Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art methods, achieving faster and more influential results (i.e., outperforming over 20%) under realistic network conditions. Moreover, our method shows robust performance in partially observable networks, effectively doubling the performance when users are predisposed to disbelieve true information.

Figures (3)

• Figure 1: CIM using Subjective Logic-based opinion model
• Figure 2: TP's influence under various CIM algorithms
• Figure 3: True party (TP)'s influence under various CIM algorithms when both TP and False party (FP) use DRL for seed node selection