PAGE-RAG: Evidence-Grounded Adaptive Graph Retrieval for Long-Document Question Answering

2026-07-21

AI Summary

Existing knowledge graphs are often treated as standalone knowledge sources, yet their incompleteness can lead to unreliable retrieval and hallucinated generations. This work proposes a projection-aware graph retrieval framework that reframes knowledge graphs as semantic scaffolds for organizing and navigating document-based knowledge, rather than substitutes for original documents. By integrating graph-augmented retrieval (GraphRAG), task-adaptive retrieval routing, explicit knowledge boundary constraints, and evidence-grounded generation, the framework dynamically selects retrieval pathways and rigorously confines responses to verifiable evidence. Experimental results demonstrate that this approach significantly enhances retrieval efficiency and knowledge reliability while maintaining high answer quality, thereby validating the efficacy of projection-aware modeling and boundary-aware control mechanisms.

Abstract

GraphRAG improves long-document question answering by introducing structured representations beyond conventional retrieval. However, automatically constructed graphs are inherently incomplete projections of source documents, and treating them as independent knowledge sources may lead to unreliable retrieval and generation. We propose PAGE-RAG, a projection-aware adaptive graph retrieval framework for reliable long-document question answering. PAGE-RAG views graph structures as semantic skeletons that organize and navigate document knowledge, rather than replacing the original knowledge source. Based on this perspective, PAGE-RAG introduces a task-adaptive retrieval routing strategy that dynamically selects appropriate retrieval behaviors according to query requirements. Furthermore, PAGE-RAG incorporates strict knowledge boundary control, ensuring that generated responses remain grounded within available evidence and abstaining from unsupported information beyond the accessible knowledge scope. Experiments demonstrate that PAGE-RAG achieves competitive answer quality while improving retrieval efficiency and knowledge reliability, highlighting the importance of projection-aware graph modeling, adaptive retrieval, and explicit knowledge boundary control for trustworthy GraphRAG systems. The source code is publicly available at https://github.com/CXY0112/PAGE-RAG.

Problem

• GraphRAG
• long-document question answering
• knowledge reliability
• retrieval grounding
• incomplete graph projection

Innovation

• projection-aware graph retrieval
• adaptive retrieval routing
• knowledge boundary control
• GraphRAG
• evidence-grounded generation