Towards Secure and Practical Crowd Wisdom Utilization

Abstract

Crowd wisdom has been extensively employed across diverse domains such as decisionmaking, problem-solving, urban planning, crisis response, crowdsourcing, and content moderation. While the versatility of crowd wisdom facilitates broad applications in practice, existing systems have stored and processed plaintext crowd data, exposing sensitive information to potential access by attackers or system administrators, thereby compromising user privacy. Consequently, it is crucial to implement strong protection for crowd data. However, achieving this protection presents challenges in effectively processing crowd data for diverse applications. This inherent tension between privacy preservation and data processing necessitates a comprehensive study into enabling secure and practical utilization of crowd wisdom.

This dissertation explores privacy-preserving designs for the use of crowd wisdom in prevalent applications. We investigate state-of-the-art privacy-preserving designs for aggregating crowd wisdom, identify key security problems in various applications, and propose feasible systems to enforce end-to-end encryption (E2EE) while using crowd data. Specifically, we first present a privacy-preserving crowdsourcing service with robust quality assurance, facilitating secure profile matching and data collection through verifiable quality reporting via custom secure protocols. Secondly, we present a privacy-friendly content moderation system based on user reporting for EEMSs. Lastly, we propose a comprehensive framework for abusive information detection that leverages a crowd-defined and transparent blocklist while preserving user privacy. This research aims to enhance the understanding of secure and practical crowd wisdom applications and to inspire the development of innovative techniques and designs within this domain.

Date of Award	24 Apr 2025
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Cong WANG (Supervisor)