VecTree-RAG: An Agentic Retrieval-Augmented Generation Framework Combining Vector and Tree Retrieval for Efficiency and Accuracy

2026-07-24

AI Summary

This work addresses the challenge in scientific question answering of simultaneously identifying relevant papers and precisely locating supporting evidence, a task where conventional approaches often fail due to their neglect of document structure, leading to disconnection between claims and context. The paper proposes the first multi-agent retrieval-augmented generation (RAG) framework that integrates dense vector retrieval with structure-aware section tree traversal: it first retrieves relevant documents at the corpus level and then employs reasoning-guided tree traversal to incrementally load necessary sections within selected papers to pinpoint evidence. Evaluated on QASPER, LitQA2, and MOSAIC, the method achieves state-of-the-art performance, with LLM-based answer accuracy ranging from 0.800 to 0.925, nearly quadrupling evidence-page precision over baselines while substantially reducing the number of tokens required for inference.

Abstract

Scientific question answering requires a retrieval system to solve two distinct problems: identifying which papers are relevant and locating the supporting evidence within those papers. Conventional retrieval-augmented generation typically addresses both through similarity search over fixed-length passages, flattening document structure and separating scientific claims from their methodological and argumentative context. We present VecTree-RAG, an agentic framework that assigns these tasks to complementary retrieval mechanisms. Vector search ranks compact document and section representations across the corpus, whereas reasoning-guided traversal of source-verified section trees localizes evidence within shortlisted papers. Full text is retained in a page store and exposed progressively only after structural localization. We evaluate VecTree-RAG on 300 QASPER questions, an open-access subset of 54 LitQA2 questions, and 49 multi-document MOSAIC questions. Compared with Dense RAG, reranked Dense RAG, RAPTOR, and Search-o1, VecTree-RAG obtained the highest observed answer score on all three benchmarks, reaching 0.800 LLM-judge correctness on QASPER, 0.925 accuracy on LitQA2, and a 0.547 composite score on MOSAIC. On QASPER, its evidence-page precision was 0.274, compared with 0.046--0.071 for the baselines. LitQA2 ablations further showed that the complete vector--tree architecture required fewer inference tokens than variants without tree navigation or corpus-level vector routing. These results indicate that vector retrieval narrows the corpus-level search space and tree navigation concentrates reading on structurally relevant evidence. Although multi-turn inference remains more expensive than single-call retrieval, VecTree-RAG provides a structure-aware and traceable architecture for scientific literature question answering.

Problem

• scientific question answering
• retrieval-augmented generation
• evidence localization
• document structure
• relevance identification

Innovation

• Vector-Tree Retrieval
• Agentic RAG
• Structure-aware QA
• Evidence Localization
• Scientific Literature Understanding