DeepWeaver: Bridging the Evidence Synthesis Gap in Open-Ended Question Answering

• URL: http://arxiv.org/abs/2608.18988v1
• Date: Wed, 19 Aug 2026 14:53:53 GMT
• Title: DeepWeaver: Bridging the Evidence Synthesis Gap in Open-Ended Question Answering
• Abstract summary: We propose DeepWeaver, a framework that weaves noisy retrieved evidence into comprehensive answers.<n>We evaluate DeepWeaver on open-ended QA over both knowledge bases and the web, and introduce LoQA, a high-density benchmark for evidence synthesis.
• Score: 49.87758530386642
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Retrieve-then-generate pipelines are commonly used to produce deep-research answers for open-ended questions, but retrieval alone is insufficient: LLMs must organize noisy and fragmented evidence into comprehensive, well-cited answers. We refer to this process as evidence synthesis. However, direct generation often underuses evidence, misaligns citations, and collapses diverse information into shallow summaries, exposing an evidence synthesis gap between retrieval and generation. Thus, we propose DeepWeaver, a novel framework that weaves noisy retrieved evidence into comprehensive answers by maintaining Thought Block Chains (TBCs), a structured representation that groups claims, salient information, keywords, and supporting evidence. DeepWeaver uses subordinate TBCs to inspect residual evidence, commit TBC revisions, and discover new claims before final generation. We evaluate DeepWeaver on open-ended QA over both knowledge bases and the web, and introduce LoQA, a high-density benchmark for evidence synthesis. Across multiple LLMs, DeepWeaver improves content sufficiency, citation grounding, and detail preservation on LoQA, while achieving deeper insights and higher citation quality on DeepResearch Bench. These results show that evidence weaving is an effective mechanism for bridging retrieval and generation in open-ended QA. Our code is available at https://github.com/KlozeWang/DeepWeaver.

Related papers

• Navigating Sparse Evidence: Agentic Visual RAG via Explicit Context Selection and Consolidation [50.92223196290564]
  We propose a unified agent loop for explicit evidence selection and consolidation.<n>During exploration, SCoRE retains only query-relevant observations and their source pointers in a maintained textual ledger.<n>At termination, it reloads the referenced original images and consolidates the visual evidence for answering, arranging it into a logical sequence.
  arXiv  Detail & Related papers  (2026-09-14T16:11:35Z)
• Beyond One-Shot Expansion: Contrastive Evidence Exploration for Multi-Hop Retrieval [19.37201042661162]
  We propose a training-free multi-hop retrieval framework that integrates evidence-conditioned exploration, passage-specific contrastive refinement, and coverage-aware final ranking.<n> Experiments on MuSiQue, HotpotQA, and 2WikiMultihopQA demonstrate consistent improvements in retrieval quality and downstream QA performance over baselines.
  arXiv  Detail & Related papers  (2026-09-07T05:13:56Z)
• A Tree-based RAG Framework for Evidence-Intensive QA via Adaptive Planning and Topology-Aware Evidence Gathering [5.690105879823579]
  Recent structured RAG methods leverage tree- or graph-based reasoning structures to improve multi-hop QA.<n>We propose APT-RAG, an Adaptive Planning and Topology-aware evidence gathering RAG framework.<n>In experiments on evidence-intensive QA benchmarks, APT-RAG outperforms existing structured RAG methods.
  arXiv  Detail & Related papers  (2026-09-04T10:34:14Z)
• SAFE-G: Structure-aware Faithful Evidence-guided Generation for Knowledge-based Visual Question Answering [8.538563236020947]
  SAFE-G is a structure-Aware Faithful Evidence-guided Generation framework.<n>It enables precise evidence localization and trustworthy reasoning.<n>It outperforms prior methods by a margin of 8.9% and 3.5%.
  arXiv  Detail & Related papers  (2026-08-22T06:34:53Z)
• Deep-Research Agents Can Be Poisoned via User-Generated Content [8.098290223993954]
  Deep-research agents issue many related queries during a single research session.<n>We show that for many common search topics, they repeatedly retrieve the same user-generated content (UGC) pages from platforms such as Reddit and Wikipedia.<n>This overlap creates a concentrated attack surface.<n>We evaluate this attack on three representative deep-research systems.
  arXiv  Detail & Related papers  (2026-05-22T21:46:32Z)
• Orchestrating Specialized Agents for Trustworthy Enterprise RAG [8.772844442593975]
  One-pass retrieval-and-write pipelines often yield shallow summaries.<n>We introduce ADORE, an agentic framework that replaces linear retrieval with iterative, user-steered investigation.<n>Our contributions are threefold: Memory-locked synthesis, Evidence-coverage-guided execution, and section-packed long-context grounding.
  arXiv  Detail & Related papers  (2026-01-26T08:48:41Z)
• WebWeaver: Structuring Web-Scale Evidence with Dynamic Outlines for Open-Ended Deep Research [73.58638285105971]
  This paper tackles textbfopen-ended deep research (OEDR), a complex challenge where AI agents must synthesize vast web-scale information into insightful reports.<n>We introduce textbfWebWeaver, a novel dual-agent framework that emulates the human research process.<n>Our framework establishes a new state-of-the-art across major OEDR benchmarks, including DeepResearch Bench, DeepConsult, and DeepResearchGym.
  arXiv  Detail & Related papers  (2025-09-16T17:57:21Z)
• DeepSieve: Information Sieving via LLM-as-a-Knowledge-Router [57.28685457991806]
  DeepSieve is an agentic RAG framework that incorporates information sieving via LLM-as-a-knowledge-router.<n>Our design emphasizes modularity, transparency, and adaptability, leveraging recent advances in agentic system design.
  arXiv  Detail & Related papers  (2025-07-29T17:55:23Z)
• NeoQA: Evidence-based Question Answering with Generated News Events [53.85274258429368]
  We introduce NeoQA, a benchmark designed to address this issue.<n>We propose our dataset as a new platform for evaluating evidence-based question answering.
  arXiv  Detail & Related papers  (2025-05-09T10:51:29Z)
• REM-Net: Recursive Erasure Memory Network for Commonsense Evidence Refinement [130.8875535449478]
  REM-Net is equipped with a module to refine the evidence by erasing the low-quality evidence that does not explain the question answering. Instead of retrieving evidence from existing knowledge bases, REM-Net leverages a pre-trained generative model to generate candidate evidence customized for the question. The results demonstrate the performance of REM-Net and show that the refined evidence is explainable.
  arXiv  Detail & Related papers  (2020-12-24T10:07:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.