Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Advisor

Zhipeng Cai

Second Advisor

Yingshu Li


In this dissertation, we investigate the privacy-preserved data publication problems towards pervasively existing linkable social-physical contents. On the one hand, data publication has been considered as a critical approach to facilitate numerous utilities for individuals, populations, platform owners, and all third-party service providers. On the other hand, the unprecedented adoption of mobile devices and the dramatic development of Internet-of-Thing (IoT) systems have pushed the collection of surrounding physical information among populations to a totally novel stage. The collected contents can provide a fine-grained access to both physical and social aspects of the crowds, which introduces a comprehensively linkable and potentially sensitive information domain. The linkage includes the related index like privacy, utility, and efficiency for sophisticated applications, the inherent correlations among multiple data sources or information dimensions, and the connections among individuals. As the linkage leads to various novel challenges for privacy preservation, there should be a body of novel mechanisms for linkable social-physical data publications.

As a result, this dissertation proposes a series of mechanisms for privacy-preserved linkable social-physical data publication. Firstly, we study the publication of physical data where the co-existing useful social proles and the sensitive physical proles of the data should be carefully maintained. Secondly, we investigate the data publication problem jointly considering the privacy preservation, data utility, and resource efficiency for task completion in crowd-sensing systems. Thirdly, we investigate the publication of private contents used for the recommendation, where contents of a user contribute to the recommendation results for others. Fourthly, we study the publications of reviews in local business service systems, where users expect to conceal their frequently visited locations while cooperatively maintain the utility of the whole system. Fifthly, we study the acquisition of privacy-preserved knowledge on cyber-physical social networks, where third-party service providers can derive the community structure without accessing the sensitive social links. We also provide detailed analysis and discussion for proposed mechanisms, and extensively validate their performance via real-world datasets. Both results demonstrate that the proposed mechanisms can properly preserve the privacy while maintaining the data utility.

At last, we also propose the future research topics to complete the whole dissertation. The first topic focuses on the privacy preservation towards correlations beneath multiple data sources. The second topic studies more privacy issues for the whole population during data publication, including both the novel threats for related communities, and the disclosure of trends within crowds.