CodaRAG：基於互補學習之關聯式知識串接檢索生成

Abstract

大型語言模型在處理知識密集型任務時，常因產生幻覺以及對分散資訊進行片段化推理而表現受限。儘管檢索增強生成（Retrieval-Augmented Generation, RAG）透過引入外部知識來支撐生成過程，現有方法多將證據視為彼此獨立的單元，難以有效重建連結這些資訊的邏輯推理鏈。受互補學習系統啟發，本文提出 CodaRAG，一個將檢索機制由被動查詢提升為主動關聯探索的框架。CodaRAG 採用三階段處理流程：（1）知識圖譜建構（knowledge graph construction）：將分散的知識抽取整合為一致且穩定的知識表示；（2）關聯性檢索（associative retrieval）：透過語意、脈絡化與功能性等多維關聯路徑，在圖結構中進行搜尋，以顯式重建分散的證據鏈；（3）干擾移除（interference elimination）：去除過度關聯所引入的雜訊，確保推理過程中的語境一致性與高精度。在 GraphRAG-Bench 上的實驗結果顯示，CodaRAG 在檢索召回率上達到 7–10% 的提升，並在生成準確率上提升 3–11%。實驗結果證明，CodaRAG 能夠在事實型、推理型、摘要型與創造型任務中，有效且系統性地提升關聯式證據檢索的穩健性與完整性。

Large Language Models (LLMs) struggle with knowledge-intensive tasks due to hallucinations and fragmented reasoning over dispersed information. While Retrieval-Augmented Generation (RAG) grounds generation in external sources, existing methods often treat evidence as isolated units, failing to reconstruct the logical chains that connect these dots. Inspired by Complementary Learning Systems (CLS), we propose CodaRAG, a framework that evolves retrieval from passive lookup into active associative discovery. CodaRAG operates via a three-stage pipeline: (1) Knowledge graph construction to unify fragmented extractions into a stable memory substrate; (2) Associative retrieval to traverse the graph via multi-dimensional pathways—semantic, contextualized, and functional—explicitly recovering dispersed evidence chains; and (3) Interference elimination to prune hyper-associative noise, ensuring a coherent, high-precision reasoning context. On GraphRAG-Bench, CodaRAG achieves absolute gains of 7-10% in retrieval recall and 3–11% in generation accuracy. These results demonstrate CodaRAG’s superior ability to robustify associative evidence retrieval for factual, reasoning, summary, and creative tasks.