DocTrace: Towards Traceable Long Document VQA via Hierarchical Evidence Graph Reasoning

2026-08-04 · Le Xiang, Zhicheng Guan, Hong Chen, Xiaocong Lin, Zhenghua Lei, Teng Hu, Bolei He, Long Zeng arxiv

Long Document Visual Question Answering (LongDocVQA) requires Multimodal Large Language Models (MLLMs) to locate, integrate, and reason over heterogeneous document elements distributed across multiple pages. Existing approaches, including end-to-end MLLMs, retrieval-augmented generation (RAG) pipelines, and document agents, often lack explicit mechanisms to represent and verify how grounded evidence is progressively composed during reasoning, limiting both answer accuracy and traceability. In this paper, we cast LongDocVQA as an explicit evidence graph reasoning problem rather than implicit answer prediction. To this end, we propose DocTrace, a hierarchical framework that progressively performs evidence localization, structured document parsing, and evidence graph reasoning to enable explicit evidence provenance. To effectively learn these capabilities, we develop a two-stage training framework: joint Supervised Fine-Tuning (SFT) first initializes evidence localization and graph reasoning abilities, followed by task-specific Group Relative Policy Optimization (GRPO) with dedicated rewards to further optimize these capabilities. Extensive experiments on MMLongBench-Doc, LongDocURL, and SlideVQA demonstrate that DocTrace consistently outperforms both existing open-source baselines and proprietary MLLMs. Compared with the Qwen3-VL-8B-Instruct backbone, DocTrace achieves absolute improvements of 14.4, 11.3, and 11.7 points on the three benchmarks, respectively. Beyond competitive performance, DocTrace constructs traceable evidence graphs with explicit node-level provenance, enabling transparent and verifiable reasoning for long document understanding.

📄 PDF Abstract BibTeX arXiv:2608.03292

Code (0)

등록된 구현이 없습니다.

Tasks

Visual Question Answering

Trace Only What You Need: Structure-Aware On-Demand Hypergraph Memory for Long-Document Question Answering

2026-06-09 · Xiangjun Zai, Xingyu Tan, Chen Chen, Xiaoyang Wang 외 arxiv

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 evi…

Question Answering

ITL: Interpretable Document Alignment with Structured Reference Frameworks

2026-08-27 · Raúl Giráldez, Dayrelis Mena, Jesús S. Aguilar--Ruiz arxiv

Measuring alignment between documents and structured reference frameworks requires identifying conceptual evidence distributed throughout the text and reporting it through measures that are quantitative, interpretable, a…

EnSI-RAG: Entity-Structure-Indexed Retrieval-Augmented Generation for Long-Document Question Answering

2026-08-21 · Xuanyu Meng, Jiashuo Sun, Jash Rajesh Parekh, Jiawei Han arxiv

Question answering (QA) over long, connected documents remains challenging because relevant evidence may span multiple entities and their relationships. Existing retrieval-augmented generation (RAG) methods typically ind…

Question Answering

SIN-Bench: Tracing Native Evidence Chains in Long-Context Multimodal Scientific Interleaved Literature

2026-01-15 · Yiming Ren, Junjie Wang, Yuxin Meng, Yihang Shi 외 arxiv

Evaluating whether multimodal large language models truly understand long-form scientific papers remains challenging: answer-only metrics and synthetic "Needle-In-A-Haystack" tests often reward answer matching without re…

Traceable Multi-Agent System for Knowledge-Based Forecasting

2026-08-04 · Junhyeok Kang, Sangjun Han, Hyeokjun Choe, Soonyoung Lee arxiv

Enterprise forecasting increasingly relies on autonomous agents that interpret documents, search for data, generate code, and revise models. While this autonomy helps build adaptive forecasting pipelines, it also makes i…