Fernando Spadea · ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 2026
DOI: 10.1145/3799682.3840232
Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).
Modern recommendation systems force a critical trade-off between user data sovereignty and algorithmic personalization, while decentralized alternatives often struggle with data sparsity and reactive interventions that fail during high-volatility events, like decentralized finance liquidations. This doctoral research outlines a comprehensive framework that bridges sovereignty-preserving collaborative learning with predictive risk management. By fusing private personal knowledge graphs with public contextual data through computationally efficient retrieval-augmented generation and federated learning, the system generates highly tailored recommendations without exposing raw user data. Concurrently, it develops proactive, agentic architectures utilizing Survival Analysis and the physics of self-organized criticality to forecast individual time-to-event risks and systemic network fragility. By integrating these macroscopic and microscopic topological signals, the framework empowers automated agents to dynamically shift from yield-maximizing recommendations to minimally invasive, protective interventions, safeguarding users against cascading failures in decentralized ecosystems.
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
No comments yet — start the discussion below.