Towards Opinion Shaping: A Deep Reinforcement Learning Approach in Bot-User Interactions

TL;DR

This work addresses shaping public opinion in social networks by leveraging bot-user interactions and targeted advertising within the Stochastic Bounded Confidence Model (SBCM). It deploys Deep Deterministic Policy Gradient (DDPG) to learn continuous-action policies for two interventions: agent-controlled bots and budget-constrained advertising. The two DRL-based schemes demonstrate measurable shifts in the mean opinion and reveal how budgets, interaction randomness, and network structure influence strategy. The study highlights the practicality of AI-driven influence in online platforms while underscoring ethical considerations and directions for more complex, multi-network and adversarial scenarios.

Abstract

This paper aims to investigate the impact of interference in social network algorithms via user-bot interactions, focusing on the Stochastic Bounded Confidence Model (SBCM). This paper explores two approaches: positioning bots controlled by agents into the network and targeted advertising under various circumstances, operating with an advertising budget. This study integrates the Deep Deterministic Policy Gradient (DDPG) algorithm and its variants to experiment with different Deep Reinforcement Learning (DRL). Finally, experimental results demonstrate that this approach can result in efficient opinion shaping, indicating its potential in deploying advertising resources on social platforms.

Figures (3)

• Figure 1: Average reward of the actor over episodes. This plot provides insights into how the agent's performance evolves over training episodes.
• Figure 2: Visualization of user and bot opinions over time steps. The plot demonstrates the dynamics of opinions and the influence of bots.
• Figure 3: The trajectory of user opinions and the influence of targeted advertising in a single episode with $\mu=0.1$, and $\epsilon$$= -2.0$.