Research Publications
My research spans from investigating algorithmic bias in social media to securing Graph Neural Networks against backdoor attacks and developing Federated Learning frameworks for privacy-preserving data analysis."
Pillar 1: Algorithmic Bias & Digital Ethics
Investigating the socio-technical impacts of recommendation algorithms on social media platforms. My work focuses on how algorithmic bias influences the digital experiences, political engagement, and social well-being of youth. I aim to develop frameworks that promote transparency and mitigate digital harms in AI-driven social ecosystems.
Pillar 2: Secure Graph Representation Learning
Advancing the robustness and security of Graph Neural Networks (GNNs). My research explores the vulnerabilities of graph-structured data to adversarial threats, specifically focusing on backdoor attacks and the development of defense mechanisms. Additionally, I work on graph generation techniques to create synthetic datasets that maintain the structural integrity and utility of real-world networks for secure AI training.
Pillar 3: Privacy-Preserving & Federated Learning
Developing decentralized machine learning frameworks that prioritize data sovereignty and user privacy. My work in Federated Learning focuses on enabling collaborative model training across distributed datasets without the need to exchange sensitive raw data. By integrating differential privacy and secure aggregation, I aim to build AI systems that are both high-performing and privacy-compliant by design.
Pillar 4: Semantic Web & Data Provenance
Building the foundational layer for trustworthy and interoperable digital information. I specialize in Linked Open Data (LOD), Knowledge Graph evolution, and Computational Trust. My research investigates how data provenance (tracking the origin and history of data) and metadata can be used to verify the quality and reliability of information in large-scale heterogeneous systems.