Jul 2026· 2026 6th International Conference on Intelligent Communications and Computing (ICICC)· pp. 506-510· 1 citation· 19 references
TL;DR
This work proposes StructAgent, a document-oriented LLM agent that integrates structure-aware navigation, sequential reading, and evidence-constrained generation into a unified agentic loop, and generates answers whose reasoning chains are explicitly grounded in cited evidence.
Abstract
Large language models (LLMs) struggle with longdocument reasoning: naively packing entire documents into the context window leads to degraded performance, while retrievalaugmented generation (RAG) fragments documents into isolated chunks that lose structural coherence. Moreover, even when existing agents produce correct answers, their intermediate reasoning steps are often unfaithful to the retrieved evidence. We propose StructAgent, a document-oriented LLM agent that integrates structure-aware navigation, sequential reading, and evidence-constrained generation into a unified agentic loop. StructAgent first parses the document hierarchy (sections, tables, and cross-references), then navigates this structure tree to locate relevant regions, reads contiguous passages to preserve local context, and finally generates answers whose reasoning chains are explicitly grounded in cited evidence. We evaluate StructAgent on three long-document benchmarks-QASPER, HotpotQA, and QuALITY-measuring both answer accuracy (F1) and reasoning faithfulness via two newly introduced metrics: Evidence-Reasoning Consistency (ERC) and Reasoning-Answer Consistency (RAC). Experimental results show that StructAgent achieves absolute F1 gains of 3.1-10.9 points depending on the baseline and dataset, along with higher faithfulness scores compared to vanilla RAG, long-context LLMs, and ReAct-based agents, especially on the length-controlled QASPER setting.
This paper proposes DocTrace, a hierarchical framework that progressively performs evidence localization, structured document parsing, and evidence graph reasoning to enable explicit evidence provenance, and develops a two-stage training framework.
Lei Xiang, Zhicheng Guan, Hong Chen et al.· 0 citations
Graph-based and multimodal retrieval frameworks provide a strong foundation for long-document question answering, but single-pass retrieval can remain brittle when queries are ambiguous, multi-step, or misaligned with the indexed evidence. We present Agentic-RAG, a structure-aware retrieval-augmented generation framework that combines a MinerU-LightRAG-based document processing and graph retrieval pipeline with an LLM-based agentic query-control layer. The base pipeline supports structure-aware parsing and VLM-based captioning for textual, visual, and tabular evidence, while the agentic layer performs query planning, LLM-based evidence reranking, answer generation, groundedness checking, relevance checking, and query reformulation. The framework does not modify the underlying graph construction or indexing mechanism; instead, it improves retrieval control by guiding the base retriever toward evidence that better matches the user's information need. Experiments on HotpotQA and ASQA show that Agentic-RAG improves context precision on HotpotQA from 0.1682 to 0.2462 and substantially improves context precision and context recall on ASQA from 0.5045/0.2783 to 0.6522/0.4783. The results indicate that agentic query control is especially useful for ambiguity-heavy long-form reasoning, while sparse multi-hop evidence chaining remains a bottleneck for future work.
D. Lam, Gia Hien Tran, Tien-Dung Do· 2026 11th International Conf...· 0 citations
REFACT is an adaptive fact-restatement citation framework that enables LLMs to determine when contextual grounding is needed and selectively restate source facts at appropriate levels of detail for reliable reasoning.
Zhensheng Jin, Xin Dai, Zhenghao Liu et al.· 0 citations
Chinese ancient document understanding demands complex visual, linguistic, and historical reasoning. Current Large Vision-Language Models (LVLMs) typically rely on an opaque, single-pass generation paradigm, often producing overconfident and weakly grounded responses. To address this, we propose SAGE, an evidence-grounded multi-agent framework that reformulates Chinese ancient document understanding as evidence-grounded inference rather than direct answer generation. SAGE coordinates specialized agents for task-aware planning, tool-mediated evidence acquisition, claim-level verification, and bounded replanning under a constrained shared-state runtime. This design supports bounded evidence seeking, answer revision, and abstention when grounding is insufficient. Experiments on the AncientDoc benchmark show that SAGE consistently outperforms matched direct-answering baselines across three LVLM backbones. Remarkably, SAGE with Qwen3.5-9B surpasses much larger monolithic LVLMs on most evaluated metrics, highlighting the importance of structured, evidence-grounded inference beyond model scaling.
Yuchuan Wu, Xuan Luo, Yinglian Zhu et al.· 0 citations
Large language model (LLM)-based search agents answer questions through multi-step interactions with external environments. However, providing complete execution trajectories to the LLM causes unbounded context growth and introduces noise. Existing compression methods reduce context at the cost of important details and often replace erroneous facts without repairing downstream reasoning derived from them. To address this problem, we propose ReTree, a self-correcting tree-structured memory mechanism for search agents. ReTree constructs a bounded per-step reasoning context while preserving source-linked evidence. It models search as an evidence tree whose nodes store bounded summaries, evidence, and revision histories. When newly retrieved evidence contradicts an earlier claim, ReTree traces back to the node where the claim was introduced, replaces outdated evidence, regenerates summaries, prunes affected branches, and resumes search. Source-grounded evidence provenance supports reliable conflict localization and keeps final claims traceable to retrieved passages. Experiments on four public question-answering and search benchmarks show that ReTree consistently outperforms Full-Trajectory ReAct, improving answer accuracy by up to 25.6 percentage points (pp); the average maximum per-step reasoning context of Full-Trajectory ReAct is $1.27$--$1.51\times$ that of ReTree. These results establish ReTree as an effective self-correcting memory abstraction for long-horizon search.
Aijun Yang, Qianxue Guo, Ziyi Huang et al.· 0 citations
Large language models (LLMs) have made substantial progress on reasoning tasks that require increasingly long and complex inferential chains. This progress primarily reflects reasoning depth. A complementary and comparatively unexamined capability is reasoning breadth: exploring multiple semantic directions in parallel and integrating the resulting clues into one coherent answer. We introduce MPAR-Bench, a bilingual English-Chinese benchmark that isolates reasoning breadth through multi-point associative reasoning. Inspired by the cooperative game Just One, each item asks a model to recover a hidden target from several independently generated, semantically diverse clues. We construct 1,000 items using a multi-agent clue-generation pipeline, embedding-based diversity filtering, and human verification. Only the answer space is drawn from public word lists, whereas every clue set is generated from scratch. Beyond exact-match accuracy, we evaluate models using accuracy, ANLS, embedding similarity, reasoning-trace verification, and four perturbations: clue masking, order shuffling, distractor injection, and multi-step clues. Across evaluated models, perturbations reduce accuracy by 9-18 percentage points in English and 5-12 percentage points in Chinese. Thinking mode improves standard-setting accuracy, especially in English, but does not consistently reduce sensitivity to perturbations. Case-level analysis also shows that extended reasoning can overturn an initially correct hypothesis. These results indicate that greater reasoning depth does not automatically confer robust reasoning breadth, and that reasoning breadth remains largely uncovered by current benchmarks.
Si'an Xie, Jiaxu Liu, Biao Yang et al.· 0 citations