TAdaRAG: Task Adaptive Retrieval-Augmented Generation via On-the-Fly Knowledge Graph Construction

2025-11-16

AI Summary

Traditional RAG suffers from context-window limitations that truncate knowledge snippets, inducing hallucinations and breaking reasoning chains; moreover, retrieving unstructured text introduces noise that impairs reasoning. To address these issues, we propose a dynamic, task-adaptive knowledge graph construction framework that integrates intent-driven routing with implicit structured knowledge extraction, enabling domain-specific, concise, and non-redundant knowledge fusion. Our method jointly optimizes retrieval and generation via supervised fine-tuning and reinforcement learning, supporting real-time construction of structured knowledge representations from external sources. Evaluated on six public benchmarks and the real-world business dataset NowNewsQA, our approach achieves significant improvements over state-of-the-art methods, demonstrating superior performance in long-context understanding, cross-domain generalization, and reasoning coherence.

Technology Category

• Knowledge Representation and Reasoning: Knowledge Acquisition
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Retrieval-Augmented Generation (RAG) improves large language models by retrieving external knowledge, often truncated into smaller chunks due to the input context window, which leads to information loss, resulting in response hallucinations and broken reasoning chains. Moreover, traditional RAG retrieves unstructured knowledge, introducing irrelevant details that hinder accurate reasoning. To address these issues, we propose TAdaRAG, a novel RAG framework for on-the-fly task-adaptive knowledge graph construction from external sources. Specifically, we design an intent-driven routing mechanism to a domain-specific extraction template, followed by supervised fine-tuning and a reinforcement learning-based implicit extraction mechanism, ensuring concise, coherent, and non-redundant knowledge integration. Evaluations on six public benchmarks and a real-world business benchmark (NowNewsQA) across three backbone models demonstrate that TAdaRAG outperforms existing methods across diverse domains and long-text tasks, highlighting its strong generalization and practical effectiveness.

Problem

• Addresses information loss from truncated knowledge chunks in RAG
• Solves irrelevant detail introduction from unstructured knowledge retrieval
• Improves reasoning chains and reduces hallucinations in LLM responses

Innovation

• On-the-fly task-adaptive knowledge graph construction
• Intent-driven routing to domain-specific extraction templates
• Reinforcement learning-based implicit extraction mechanism