Privacy-Preserving Data Publishing is a research thesis by Benjamin C. M. Fung, focusing on protecting individual privacy while enabling data mining. The thesis addresses privacy threats in data publishing, such as linking individuals to sensitive information through quasi-identifiers. It proposes a unified solution to anonymize data while preserving its usefulness for data mining. The research explores various models of data publishing, including single releases, sequential releases, and secure data integration. It introduces a framework called Top-Down Refinement (TDR) for anonymizing classification data, and extends this to handle confidence bounding, sequential anonymization, and secure data integration. The thesis also compares privacy-preserving data publishing (PPDP) with privacy-preserving data mining (PPDM), highlighting their differences and focusing on PPDP's role in protecting privacy during data publishing. Key contributions include developing algorithms for anonymizing data, bounding confidence in inferences, and ensuring secure data integration across multiple publishers. The research emphasizes the importance of balancing privacy protection with data utility, and provides practical solutions for real-world data publishing scenarios. The thesis is approved by a committee of experts in computer science and information security, and includes detailed experiments and evaluations of the proposed methods.Privacy-Preserving Data Publishing is a research thesis by Benjamin C. M. Fung, focusing on protecting individual privacy while enabling data mining. The thesis addresses privacy threats in data publishing, such as linking individuals to sensitive information through quasi-identifiers. It proposes a unified solution to anonymize data while preserving its usefulness for data mining. The research explores various models of data publishing, including single releases, sequential releases, and secure data integration. It introduces a framework called Top-Down Refinement (TDR) for anonymizing classification data, and extends this to handle confidence bounding, sequential anonymization, and secure data integration. The thesis also compares privacy-preserving data publishing (PPDP) with privacy-preserving data mining (PPDM), highlighting their differences and focusing on PPDP's role in protecting privacy during data publishing. Key contributions include developing algorithms for anonymizing data, bounding confidence in inferences, and ensuring secure data integration across multiple publishers. The research emphasizes the importance of balancing privacy protection with data utility, and provides practical solutions for real-world data publishing scenarios. The thesis is approved by a committee of experts in computer science and information security, and includes detailed experiments and evaluations of the proposed methods.