CodaRAG: Connecting the Dots with Associativity Inspired by Complementary Learning

2026-04-11

AI Summary

This work addresses the tendency of large language models to hallucinate in knowledge-intensive tasks and their difficulty in performing coherent reasoning over fragmented information. While existing retrieval-augmented generation (RAG) approaches treat retrieved evidence as isolated units lacking logical connections, this paper draws inspiration from complementary learning systems to propose CodaRAG—a novel framework that integrates complementary learning mechanisms into RAG for the first time. CodaRAG constructs a graph-structured memory with active association capabilities through a three-stage pipeline: knowledge integration, multidimensional relational graph traversal, and interference elimination, explicitly reconstructing evidence chains across semantic, contextual, and functional dimensions. Evaluated on GraphRAG-Bench, the method improves retrieval recall by 7–10% and generation accuracy by 3–11%, significantly enhancing performance on factuality, reasoning, and creative tasks.

Technology Category

• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

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 Consolidation to unify fragmented extractions into a stable memory substrate; (2) Associative Navigation 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 systematically robustify associative evidence retrieval for factual, reasoning, and creative tasks.

Problem

• Knowledge-intensive tasks
• Hallucinations
• Fragmented reasoning
• Retrieval-Augmented Generation
• Logical chains

Innovation

• Associative Retrieval
• Complementary Learning Systems
• Knowledge Graph Navigation
• Retrieval-Augmented Generation
• Interference Elimination