Document QA

QA: Question Answering

Document question answering studies how models find and combine evidence in sources such as PDFs, reports, forms, and document collections. Common approaches include retrieval-augmented generation, layout-aware and multimodal models, hierarchical indexes, and agents that navigate pages or inspect selected regions. Benchmarks increasingly test long-context and multi-document reasoning, numerical and temporal questions, and whether answers are supported by precise citations or visual evidence. Models still struggle to gather dispersed evidence and interpret complex layouts, while systems trade accuracy against latency, cost, and reliability.

Latest papers

93 papers12 papers in the last four weeks, up 50% on the four weeks before. 0.1% of all new papers.

Office Comprehension Benchmark
May 29, 2026Firoz Shaik, Mateus Picanço Lima Gomes, Tanvir Aumi +17Document Understanding LLM Evaluation
Microsoft · † Work done while at Microsoft.
HiKEY: Hierarchical Multimodal Retrieval for Open-Domain Document Question Answering
May 28, 2026Joongmin Shin, Gyuho Shim, Jeongbae Park +2Multimodal RAG Multimodal IR
Human-inspired AI Research, Korea University · Department of Computer Science and Engineering, Korea University · Computer Science and Engineering, Konkuk University
ACL-Verbatim: hallucination-free question answering for research
May 20, 2026Gábor Recski, Szilveszter Tóth, Nadia Verdha +2Hallucination in Language Models Retrieval-Augmented Generation
TU Wien · KR Labs
CiteVQA: Benchmarking Evidence Attribution for Trustworthy Document Intelligence
May 13, 2026Dongsheng Ma, Jiayu Li, Zhengren Wang +9VLM Evaluation Citation Verification
Peking University · Shanghai Artificial Intelligence Laboratory
Qwen Goes Brrr: Off-the-Shelf RAG for Ukrainian Multi-Domain Document Understanding
May 11, 2026Anton Bazdyrev, Ivan Bashtovyi, Ivan Havlytskyi +2Retrieval-Augmented Generation LLM Reranking
National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute"
ASTRA-QA: A Benchmark for Abstract Question Answering over Documents
May 11, 2026Shu Wang, Shansong Zhou, Xinyang Wang +3Question Answering Document QA
The Chinese University of Hong Kong, Shenzhen · Data Science Group, Huolala
DocScope: Benchmarking Verifiable Reasoning for Trustworthy Long-Document Understanding
May 9, 2026Xiang Feng, Jiawei Zhou, Zhangfeng Huang +6Multimodal Understanding LLM Answer Verification
School of Computer Science, National Engineering Research Center for Multimedia Software and Hubei Key Laboratory of Multimedia and Network Communication Engineering, Wuhan University, China · Alibaba Group, Hangzhou, China · Independent Researcher +1
Towards Self-Explainable Document Visual Question Answering with Chain-of-Explanation Predictions
May 7, 2026Kjetil Indrehus, Adrian Duric, Changkyu Choi +1Vision-Language Grounding Document QA
Department of Informatics, University of Oslo · Integreat – Norwegian Centre for Knowledge-driven Machine Learning · TRUST – The Norwegian Centre for Trustworthy AI
DocPrune:Efficient Document Question Answering via Background, Question, and Comprehension-aware Token Pruning
Apr 24, 2026Joonmyung Choi, Sanghyeok Lee, Jongha Kim +4Efficient VLM Inference Document QA
Korea University · KAIST · Adobe Research
Navigating Large-Scale Document Collections: MuDABench for Multi-Document Analytical QA
Apr 24, 2026Zhanli Li, Yixuan Cao, Lvzhou Luo +1Multi-Document QA Document QA
State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences (CAS), Beijing 100190, China · University of Chinese Academy of Sciences, Beijing 100049, China
An Answer is just the Start: Related Insight Generation for Open-Ended Document-Grounded QA
Apr 21, 2026Saransh Sharma, Pritika Ramu, Aparna Garimella +1Scientific QA Document QA
Adobe Research, India · University of Maryland, College Park
Toward Agentic RAG for Ukrainian
Apr 16, 2026Marta Sumyk, Oleksandr KosovanAgentic RAG Document QA
Ukrainian Catholic University, Lviv, Ukraine
Doc-V*:Coarse-to-Fine Interactive Visual Reasoning for Multi-Page Document VQA
Apr 15, 2026Yuanlei Zheng, Pei Fu, Hang Li +9Document QA Long-Document VQA
School of Software Engineering, Huazhong University of Science and Technology · MiLM Plus, Xiaomi Inc. · School of Computer Science and Technology, Huazhong University of Science and Technology +1
NaviRAG: Towards Active Knowledge Navigation for Retrieval-Augmented Generation
Apr 14, 2026Jihao Dai, Dingjun Wu, Yuxuan Chen +4Retrieval-Augmented Generation Agentic RAG
Tsinghua University · Nanjing University · Northeastern University
VISOR: Agentic Visual Retrieval-Augmented Generation via Iterative Search and Over-horizon Reasoning
Apr 10, 2026Yucheng Shen, Jiulong Wu, Jizhou Huang +3Agentic RAG Multimodal RAG
Soochow University, Suzhou, Jiangsu, China · Baidu Inc., Beijing, China
MAB-DQA: Addressing Query Aspect Importance in Document Question Answering with Multi-Armed Bandits
Apr 10, 2026Yixin Xiang, Yunshan Ma, Xiaoyu Du +3Document QA Multimodal RAG
Nanjing University of Science and Technology · Singapore Management University · Nanjing Pami Intelligent Technology Co., Ltd. +2
GROUNDEDKG-RAG: Grounded Knowledge Graph Index for Long-document Question Answering
Apr 6, 2026Tianyi Zhang, Andreas MarfurtQuestion Answering Document QA
getAbstract · Lucerne University of Applied Sciences and Arts Switzerland
Long-Document QA with Chain-of-Structured-Thought and Fine-Tuned SLMs
Mar 31, 2026Zhuowen Liang, Xiaotian Lin, Zhengxuan Zhang +3Question Answering Document QA
The Hong Kong University of Science and Technology (Guangzhou) · EvenUp, USA
From PDF to RAG-Ready: Evaluating Document Conversion Frameworks for Domain-Specific Question Answering
Mar 30, 2026José Guilherme Marques dos Santos, Ricardo Yang, Rui Humberto Pereira +8Document Parsing RAG Evaluation
Faculty of Engineering, University of Porto, 4200-465 Porto, Portugal · Department of Business Administration, University of Maia, 4475-690 Maia, Portugal · LIACC—Artificial Intelligence and Computer Science Laboratory, University of Porto, 4200-465 Porto, Portugal +3
EuraGovExam: A Multilingual Multimodal Benchmark from Real-World Civil Service Exams
Mar 28, 2026Jaeseong Kim, Chaehwan Lim, Sang Hyun Gil +1Multilingual VLM Evaluation Multimodal QA
School of Computer Science / Data Intelligence Lab, Semyung University, Jecheon, Republic of Korea
Chronological Knowledge Retrieval: A Retrieval-Augmented Generation Approach to Construction Project Documentation
Mar 25, 2026Ioannis-Aris Kostis, Natalia Sanchiz, Steeve De Schryver +2Document QA Temporal QA
Université Catholique de Louvain, ICTEAM/INGI, Place Sainte Barbe 2, Louvain-la-Neuve, 1348, Belgium · Buildwise, Avenue P. Holoﬀe 21, Limelette, 1342, Belgium · BPC Group, Avenue Edmond Van Nieuwenhuyse 30, Brussels, 1160, Belgium
FDARxBench: Benchmarking Regulatory and Clinical Reasoning on FDA Generic Drug Assessment
Mar 20, 2026Betty Xiong, Jillian Fisher, Benjamin Newman +5Document QA Medical QA
Stanford University · University of Washington · U.S. Food and Drug Administration
ESG-Bench: Benchmarking Long-Context ESG Reports for Hallucination Mitigation
Mar 13, 2026Siqi Sun, Ben Peng Wu, Mali Jin +3LLM Hallucination Mitigation LLM Safety Benchmarks
School of Computer Science, University of Sheffield, Sheffield, UK
Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections
Mar 12, 2026Łukasz Borchmann, Jordy Van Landeghem, Michał Turski +12Multi-Document QA LLM Agent Evaluation
Citation-Enforced RAG for Fiscal Document Intelligence: Cited, Explainable Knowledge Retrieval in Tax Compliance
Mar 11, 2026Akhil Chandra ShanivendraDocument QA Knowledge-Intensive QA
Independent Researcher
OfficeQA Pro: An Enterprise Benchmark for End-to-End Grounded Reasoning
Mar 9, 2026Krista Opsahl-Ong, Arnav Singhvi, Jasmine Collins +10Multi-Document QA LLM Agent Evaluation
Databricks
SPD-RAG: Sub-Agent Per Document Retrieval-Augmented Generation
Mar 9, 2026Yagiz Can Akay, Muhammed Yusuf Kartal, Esra Alparslan +2Multi-Document QA Retrieval-Augmented Generation
TOBB University of Economics and Technology · OSTIM Technical University
InfoGatherer: Principled Information Seeking via Evidence Retrieval and Strategic Questioning
Mar 6, 2026Maksym Taranukhin, Shuyue Stella Li, Evangelos Milios +3Document QA LLM Decision-Making
University of British Columbia · Vector Institute · University of Washington +2
MoDora: Tree-Based Semi-Structured Document Analysis System
Feb 26, 2026Bangrui Xu, Qihang Yao, Zirui Tang +8Document QA Document Layout Analysis
Shanghai Jiao Tong University · Microsoft Research · Beihang University +2
GATES: Self-Distillation under Privileged Context with Consensus Gating
Feb 24, 2026Alex Stein, Furong Huang, Tom GoldsteinKnowledge Distillation On-Policy Self-Distillation
University of Maryland, College Park

May 29, 2026J/K move · Enter open · S save

Office Comprehension Benchmark

Firoz Shaik, Mateus Picanço Lima Gomes, Tanvir Aumi, Jingci Wang+16

Microsoft · † Work done while at Microsoft.

Document Understanding LLM Evaluation Document QA Knowledge-Intensive QA

Open paper AskPDF arXiv

We introduce Office Comprehension Bench (OCB), the first public benchmark to jointly evaluate LLM systems on Word, Excel, and PowerPoint comprehension over native file formats (.docx, .xlsx, .pptx) and their variants. OCB consists of two tracks. File Fidelity Q&A tests structural and visual perception of office artifacts - tables, charts, embedded images, formulas, and app-specific elements such as headers, speaker notes, and named ranges. Domain Q&A tests expert-level reasoning grounded in real-world industry documents across 12 professional domains, with queries requiring multi-step analysis and synthesis across documents. Each reference answer is decomposed into atomic, binary-gradable claims, and an ensemble of LLM judges scores responses against each claim independently. Even the strongest frontier system in its default reasoning mode reaches only about 59.3% on Domain Q&A; increasing thinking depth within a tier does not move performance materially, while moving to a higher product tier yields modest gains. We release the dataset, evaluation tooling, judge prompt, and a public leaderboard.