Solving Freshness in RAG: A Simple Recency Prior and the Limits of Heuristic Trend Detection

2025/09/20 by Grofsky, Matthew · 1 citation
#Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · doi:10.48550/arxiv.2509.19376

Abstract

We address temporal failures in RAG systems using two methods on cybersecurity data. A simple recency prior achieved an accuracy of 1.00 on freshness tasks. In contrast, a clustering heuristic for topic evolution failed (0.08 F1-score), showing trend detection requires methods beyond simple heuristics.

Citations

• Reading Between the Timelines: RAG for Answering Diachronic Questions

Cited by

• Reliable Post-Retrieval Assembly for Agent Memory: Separating Evidence Extraction from Policy Execution

Related

• Learn to Adapt: Robust Drift Detection in Security Domain
• On the Reliable Detection of Concept Drift from Streaming Unlabeled Data
• Are Concept Drift Detectors Reliable Alarming Systems? -- A Comparative Study
• MithraDetective: A System for Cherry-picked Trendlines Detection
• Learn over Past, Evolve for Future: Forecasting Temporal Trends for Fake News Detection