RAG-Match: Retrieval-Augmented Knowledge Injection and Hierarchical Reasoning for Calibrated Semantic Relevance
Improves relevance ranking for knowledge-intensive search tasks by addressing implicit intent and fine-grained distinctions.
RAG-Match introduces a three-stage framework combining knowledge-augmented pretraining, hierarchical reasoning alignment, and preference-based calibration for semantic relevance judgment, outperforming LLM baselines on a real-world search benchmark.
Semantic relevance judgment for search is particularly challenging in knowledge-intensive scenarios, where accurate ranking requires not only semantic matching but also background grounding, multi-step reasoning, and well-calibrated decision boundaries. Existing relevance models mainly rely on direct label supervision or shallow semantic similarity, which limits their ability to handle implicit intent, factual equivalence, and fine-grained relevance distinctions. To address this issue, we propose \textsc{RAG-Match}, a three-stage framework that integrates knowledge-augmented pretraining, hierarchical reasoning alignment, and preference-based decision calibration for relevance modeling. The key idea is to first strengthen query-centered semantic grounding, then align the model with structured relevance reasoning, and finally correct decision-level inconsistencies in difficult boundary cases. Experimental results on a real-world search relevance benchmark show that \textsc{RAG-Match} consistently outperforms strong LLM-based baselines across multiple ranking metrics, demonstrating the effectiveness of combining knowledge injection, reasoning supervision, and preference optimization for fine-grained relevance judgment.