How to promote sustainable recycling of plastic packaging waste? A study combining machine learning with gaming theory

Abstract

Multi-stakeholder participation is essential for the sustainable development of plastic packaging waste recycling (PPWR), particularly given the dynamic influence of social media on behavior. This study employed machine learning techniques to analyze emotional orientations toward PPWR. Based on this analysis, a multi-agent model framework was developed to investigate behavioral interactions and evaluate the impacts of environmental communication, image signaling, and regulatory measures. Sentiment analysis indicated that public emotions toward PPWR were predominantly neutral, suggesting a need for enhanced strategies to foster sustainable recycling behavior. Findings from the multi-agent game analysis demonstrated that social media can positively affect the behavior of various stakeholders within the PPWR system. Environmental communication was found to positively influence the actions of implementing agents, while image signaling effects were observed to positively impact their evolutionary trajectories but had minimal effect on the rate of behavior change among residents. Regulatory measures were shown to significantly affect both the evolutionary trajectories and convergence rates of stakeholders. These results offer insights into predicting the integrated steady-state behaviors of multiple stakeholders in PPWR and suggest strategies for promoting sustainable recycling practices.