This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
This paper proposes H-NeiFi, a novel methodology for efficiently reaching global consensus in the opinion formation process of social media. It points out the problems of existing opinion intervention methods that hinder user autonomy and cause resistance, and H-NeiFi indirectly regulates user interactions through a two-layer dynamic model that considers the roles of experts and non-experts and a non-invasive neighbor filtering method. It efficiently controls the information propagation path by optimizing the long-term reward function using multi-agent reinforcement learning (MARL), and does not directly intervene in user interactions. Experimental results show that H-NeiFi improves the consensus speed by 22.0% to 30.7% and maintains global convergence without experts. This suggests a new social network governance paradigm that induces natural and efficient consensus while protecting user interaction autonomy.
Takeaways, Limitations
•
Takeaways:
◦
Presenting a non-invasive method of eliciting feedback that respects user autonomy
◦
Proving that global consensus is possible even without experts
◦
Optimizing efficient information propagation paths through multi-agent reinforcement learning
◦
Presenting a new paradigm for social media governance
◦
Experimental results showing that consensus speed is improved by 22.0% to 30.7%
•
Limitations:
◦
Further verification of the applicability of the proposed model to real social networks is needed.
◦
Generalizability studies across different types of social media platforms and user characteristics are needed
◦
Consideration of social impacts and ethical issues in the long term is needed.
◦
Need to review the accuracy and objectivity of social role classification in the two-tier dynamic model
