PGVMatch: Privacy-Preserving and Fine-Grained Crowdsourcing Task Matching With Lightweight On-Chain Public Verifiability

Secure task matching has been a crucial research problem in crowdsourcing, requiring the alignment of workers’ preferences and requesters’ task requirements while ensuring user privacy and matching integrity. Recently, some researchers applied blockchain to crowdsourcing, either replacing the platform for decentralization or recording proofs for public verification to defend against malicious platforms. However, they still suffer from unitary coarse-grained matching models or expensive on-chain costs. To address these limitations, we propose PGVMatch, a privacy-aware and fine-grained crowdsourcing task-matching scheme with lightweight on-chain public verifiability. Our scheme is constructed on our newly proposed cryptographic primitive–Multi-authority Attribute-Based Keyword Search with Public Verifiability (MABKS-PV), which avoids access policy leakage and key escrow risks on a single authority, meanwhile adding constant-size proof generation and lightweight verification algorithms to a basic ABKS construction. In PGVMatch, requesters can select workers with fine-grained attribute demands, and workers can pick interested tasks with multi-keyword search, preserving dual-side privacy. The matching process is conducted off-chain, while constant-size proofs are recorded on-chain for efficient and public verification of matching integrity. Security analysis and extensive experiments on the Hyperledger Fabric blockchain demonstrate both the security and our superior performance. PGVMatch outperforms the existing scheme with the fastest matching result verification, achieving a 29% improvement in throughput and a 33% reduction in latency.