AI RESEARCH

Ranking Free RAG: Replacing Re-ranking with Selection in RAG for Sensitive Domains

arXiv CS.CL

ArXi:2505.16014v5 Announce Type: replace Retrieval-Augmented Generation (RAG) systems deployed in sensitive domains must provide interpretable evidence selection and robust safeguards against data poisoning, yet current approaches rely on opaque similarity-based retrieval with arbitrary top-k cutoffs that offer no explanation for their selections and remain vulnerable to adversarial manipulation.