Task-Driven Dynamic Metadata Mapping for Scalable and Privacy-Aware Blockchain-Based Content Governance

Abstract

Blockchain-based systems ensures data immutability and traceability, making them well suited for decentralized content governance. However, conventional metadata anchoring strategies remain static and task-agnostic, resulting in suboptimal trade-offs between auditability, privacy, and scalability. To address this limitation, we propose a Task-Driven Dynamic Metadata Mapping (TDMM) mechanism that adapts anchoring strategies based on task semantics and user roles. TDMM classifies metadata into three distinct anchoring types: full on-chain anchoring for audit-critical tasks, selective disclosure via zero-knowledge proofs (ZKPs) for sensitive attributes, and off-chain reference anchoring for general-purpose metadata. A dedicated mapping controller dynamically routes metadata fields to the appropriate anchoring mode according to predefined task policies. To preserve privacy without sacrificing verifiability, TDMM incorporates a hybrid ZKP architecture that processes proofs off-chain while anchoring verification results on-chain. We implement TDMM on a Hyperledger Fabric network, augmented with IPFS and Circom-based ZKP tooling. Experimental results show that TDMM significantly reduces on-chain storage overhead, lowers re-identification risk, and supports task-appropriate latency and throughput trade-offs, demonstrating its effectiveness in balancing transparency, privacy, and scalability in decentralized metadata governance.