Enabling Fuzzy Matching in Privacy-preserving Bilateral Task Recommendation

Privacy-preserving bilateral task recommendation is an emerging functionality in mobile crowdsensing, enabling two-sided matching for both task publishers and workers while ensuring the confidentiality of both participants' attributes and task-matching policies. However, existing approaches rarely integrate fuzzy matching into privacy-preserving bilateral task recommendations, limiting the efficiency of two-sided matching because they can hardly find intersections between matched task publishers and matched workers. Intuitively leveraging fuzzy matching leaks whether the attributes of a participant satisfy a task recommendation policy. Motivated by this, this paper proposes PBTR, the first privacy-preserving bilateral task recommendation scheme that supports fuzzy matching while preserving matching privacy. Specifically, we leverage the Lagrange interpolation theorem-based secret sharing to enrich the matchmaking encryption technique to achieve fuzzy matching in bilateral task recommendations. To preserve matching privacy, we design a privacy-preserving matching proxy mechanism, where policy matches successfully if and only if the policies of both parties are satisfied. Formal security analysis proves the security of PBTR under the chosen-plaintext attack in the random oracle model. Experimental results demonstrate the feasibility of PBTR, which only requires several milliseconds to achieve privacy-preserving bilateral task recommendations.