Evaluate and Guard the Wisdom of Crowds: Zero Knowledge Proofs for Crowdsourcing Truth Inference

TL;DR

This work introduces zero-knowledge truth inference (zkTI), a protocol that makes crowdsourcing-based truth inference verifiable without exposing workers' responses. By converting truth-inference algorithms into circuits and embedding them within zkSNARK backends, zkTI guarantees correct aggregation and fair worker evaluation while preserving privacy. The authors instantiate two algorithms (CRH and ZC), develop generic decimal-arithmetic circuits, and provide a proof-of-concept implementation showing favorable efficiency against prior verifiable computing approaches, with real-data accuracy gains over simple majority voting. The approach has broad applicability to data annotation, decentralized blockchain oracles, and other domains requiring high-precision, verifiable aggregation under privacy constraints.

Abstract

Crowdsourcing has emerged as a prevalent method for mitigating the risks of correctness and security in outsourced cloud computing. This process involves an aggregator distributing tasks, collecting responses, and aggregating outcomes from multiple data sources. Such an approach harnesses the wisdom of crowds to accomplish complex tasks, enhancing the accuracy of task completion while diminishing the risks associated with the malicious actions of any single entity. However, a critical question arises: How can we ensure that the aggregator performs its role honestly and each contributor's input is fairly evaluated? In response to this challenge, we introduce a novel protocol termed $\mathsf{zkTI}. This scheme guarantees both the honest execution of the aggregation process by the aggregator and the fair evaluation of each data source. It innovatively integrates a cryptographic construct known as zero-knowledge proof with a category of truth inference algorithms for the first time. Under this protocol, the aggregation operates with both correctness and verifiability, while ensuring fair assessment of data source reliability. Experimental results demonstrate the protocol's efficiency and robustness, making it a viable and effective solution in crowdsourcing and cloud computing.

Figures (4)

• Figure 1: Illustration for two cases of crowdsourcing. In (a) $\mathcal{D}$ crowdsources his problems through a middle entity $\mathcal{A}$, $\mathcal{A}$ is responsible for the aggregation process. In (b) $\mathcal{D}$ crowdsources the problems to workers directly. It acts both as the data owner and the aggregator. After this process, anyone who doubts about the result can act as a verifier using our zkTI protocol to check the correctness of the result.
• Figure 2: The complete workflow of the zkTI protocol in our system. Upon the foundation of the crowdsourcing process, $\mathcal{A}$ proves the correctness of the algorithm and the consistency of the inputs, allowing any entity to verify it.
• Figure 3: Prover time (aggregator overhead) comparison with ASIACCS:XLXRZSD20 applying CRH. (a) $n=25$, with the different number of workers. (b) $m=75$, with the different number of tasks.
• Figure 4: Performance of algorithms under different backends varying dataset size. (a) Groth16 for ZC (b) Spartan for ZC (c) Groth16 for CRH (d) Spartan for CRH. In each subfigure we measure the prover time (aggregator overhead), verifier time (verification overhead) and proof size (communication overhead). In (a) (c), the verifier time and the proof size are tiny, both laying at the bottom of the subfigure.

Theorems & Definitions (5)

• Lemma 1
• Lemma 2
• Lemma 3
• Theorem 1
• Definition 1