Differentially Private Release of Count-Weighted Graphs

This work proposes many contributions to privacy in complex systems, mainly ones modeled as count-weighted graphs. As graph data usually contain users’ sensitive information, preserving privacy when releasing this type of data becomes a crucial issue. In this context, differential privacy (DP) has become the de facto standard for data release under strong mathematical guarantees. However, various challenges persist in effectively implementing DP to graph data, including balancing privacy protection with data utility and scalability concerns. To bridge these gaps, we propose several efficient techniques and approaches to release graph data while maintaining a robust level of privacy protection. Our results were published in the top-tier venues in the field of data management. Additionally, we disseminated our knowledge and expertise obtained during this Ph.D. research through tutorials and short courses presented at both national and international conferences.