analiza aiqbench
DocTrace introduces a multi-agent RAG framework that leverages hypergraph memory and document structure to significantly improve long-document question answering performance while cutting computational costs by over 50%. By enabling query-triggered knowledge organization and reusing historical reasoning plans, the method effectively addresses key limitations in current structured retrieval systems. This approach offers a practical path forward for scaling LLM reasoning over complex, lengthy texts without incurring prohibitive resource overhead.
Trace Only What You Need: Structure-Aware On-Demand Hypergraph Memory for Long-Document Question Answering
arXiv:2606.10921v1 Announce Type: new Abstract: Long-document question answering (QA) requires large language models (LLMs) to reason over evidence scattered across lengthy documents, where answers often depend on event order, section-level context, and cross-part evidence connections. Although retrieval-augmented generation (RAG) reduces the input context by retrieving relevant evidence, existing structured RAG methods still face three limitations: costly query-agnostic knowledge organization, insufficient use of original document structure, and no reuse of historical reasoning experience. To address these limitations, we propose DocTrace, a multi-agent RAG framework for long-document QA that supports query-triggered knowledge organization, document-structure-aware and experience-guided reasoning. DocTrace preserves document hierarchy with a lightweight document structural tree index, constructs agent-shared hypergraph-structured working memory on demand during reasoning, and stores successful reasoning plans in graph-structured experience memory for future reuse, enabling adaptive exploration across related long-document questions. Experiments on four long-document QA datasets show that DocTrace achieves the best performance on three datasets, surpassing the strongest baseline, ComoRAG, by up to 8.85% in F1 and 4.40% in EM, while reducing the overall computational cost by 53.32%
Abstract:Long-document question answering (QA) requires large language models (LLMs) to reason over evidence scattered across lengthy documents, where answers often depend on event order, section-level context, and cross-part evidence connections. Although retrieval-augmented generation (RAG) reduces the input context by retrieving relevant evidence, existing structured RAG methods still face three limitations: costly query-agnostic knowledge organization, insufficient use of original document structure, and no reuse of historical reasoning experience. To address these limitations, we propose DocTrace, a multi-agent RAG framework for long-document QA that supports query-triggered knowledge organization, document-structure-aware and experience-guided reasoning. DocTrace preserves document hierarchy with a lightweight document structural tree index, constructs agent-shared hypergraph-structured working memory on demand during reasoning, and stores successful reasoning plans in graph-structured experience memory for future reuse, enabling adaptive exploration across related long-document questions. Experiments on four long-document QA datasets show that DocTrace achieves the best performance on three datasets, surpassing the strongest baseline, ComoRAG, by up to 8.85% in F1 and 4.40% in EM, while reducing the overall computational cost by 53.32%
Computation and Language (cs.CL)
arXiv:2606.10921 [cs.CL]
(or arXiv:2606.10921v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.10921 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 263 cuvinte
Subiecte conexe
Trace Only What You Need: Structure-Aware On-Demand Hypergraph Memory for Long-Document Question Answering
arXiv · 7 aug. 2026
Salesforce after Dreamforce: How the CRM giant can grow beyond its own interface
SiliconANGLE · 19 sept. 2026
BRICS Draws a New Global Line: Full Text of New Delhi Declaration
slguardian.org · 12 sept. 2026
Models Can't Remember Their Training. Neither Can You.
LessWrong · 19 iul. 2026