A Machine Learning Augmented Cooperative-Game Framework for Blockchain and Non-Fungible Token-Based Artwork Trading...

Area of Science:

• Computer Science
• Artificial Intelligence
• Blockchain Technology

Background:

• Non-Fungible Tokens (NFTs) are revolutionizing digital art markets with secure, decentralized transactions.
• Integrating Machine Learning (ML) into NFT trading is essential for adaptability and intelligence.
• Existing Cooperative Game Theoretic Trading (CoGTT) frameworks have limitations in ML utilization across trading phases.

Purpose of the Study:

• To address gaps in real-time adaptability, negotiation strategies, and buyer-seller matchmaking in NFT trading.
• To integrate ML into a three-phase CoGTT framework for enhanced decision-making and pricing in NFT markets.
• To develop scalable, privacy-compliant digital marketplaces for smart cities using ML-driven NFT trading.

Main Methods:

• Implemented ML algorithms (decision trees, clustering, reinforcement learning - Q-learning) within a public blockchain simulation.
• Utilized smart contracts and a customized dataset reflecting market dynamics and artist credibility for simulation.
• Employed Zero-knowledge Proofs (ZKPs) for privacy preservation in transactions.

Main Results:

• Random Forest model achieved high real-time NFT price prediction accuracy (R² = 0.9920).
• K-Means clustering effectively segmented market participants for targeted negotiation (silhouette score = 0.8178).
• Q-learning integrated with Random Forest enabled dynamic bidding strategies, minimizing price discrepancies.

Conclusions:

• ML-driven NFT trading systems can significantly enhance decision-making, pricing, and adaptability.
• The proposed framework supports scalable and privacy-compliant digital marketplaces within smart city contexts.
• Automated, data-driven processes align NFT trading behavior with dynamic market demands.