Social Knowledge for Cross-Domain User Preference Modeling

This collection of recent arXiv papers contains multiple findings with clear security and privacy implications. Key points: (1) "Social Knowledge for Cross-Domain User Preference Modeling" demonstrates that large-scale social embeddings (derived from X/Twitter) can project users into a joint social space enabling accurate zero‑shot prediction of preferences across domains — a capability that facilitates robust cross‑domain profiling, targeted recommendations/ads, and scalable user modeling when combined with LLMs. (2) The Community Notes study shows an algorithmic design that systematically "u