What happened
arXiv:2608.12335v1 Announce Type: new Abstract: Financial question answering over annual reports requires more than retrieving semantically similar passages. It often involves identifying relevant companies and fiscal y
Analysis and impact
导读摘要
正文
arXiv 于 2026 年 8 月 14 日发布了一篇题为《HC-RAG: Evidence-Centric Retrieval-Augmented Generation over Heterogeneous Financial Filings》的论文。根据摘要，该研究指出金融年报问答任务面临的核心挑战并非简单的语义相似段落检索——它往往需要先识别相关公司及其财年信息，再据此定位证据。摘要原文在"fiscal y"处截断，后续细节原文未提供。
从论文标题可以推断，HC-RAG 的核心思路是"以证据为中心"（Evidence-Centric），即在异构金融申报文件（如年报、季报、监管披露等不同格式与结构的文档）之上构建检索增强生成流程。这与传统 RAG 直接按语义相似度召回文本段落的做法形成对比，暗示其可能在检索前引入了公司实体识别与财年对齐等前置步骤。不过，由于摘要不完整，具体的方法架构、实验设置与性能数据均原文未提供。
同日 arXiv cs.CL 分类下还发布了多篇相关方向的论文，共同勾勒出当前大语言模型推理与合规性研究的几个活跃分支。其中一篇关于"语用约束推理中的激活瓶颈"的研究发现，当显性表面线索与隐性可行性约束竞争时，大模型往往失败，且聚合准确率可能掩盖真实的约束推理缺陷（arXiv）。另一篇关于武装冲突预测中自反思机制的研究，通过受控消融实验考察不确定性路由的贡献（arXiv）。还有研究探讨 AI 智能体为何违反规则，发现明确设定惩罚反而可能将法律义务转化为成本收益计算，从而助长违规行为（arXiv）。
这些同期工作虽与 HC-RAG 主题不同，但共享一个底层关切：大模型在真实任务中的表现不仅取决于检索或生成能力本身，还受到约束理解、推理路径与任务框架设计的影响。对于金融问答这类高精度场景，HC-RAG 提出的"证据中心"取向或许正是对这一关切的回应——先锚定证据来源的可靠性，再执行生成。后续完整论文的发布将有助于判断该框架的实际效果。
References
- arXiv · cs.CL ↗
- LLMs Know the Constraint But Do Not Use It: Activation Bottlenecks in Pragmatic Constraint Reasoning ↗
- What Drives LLM Self-Reflection? A Controlled Ablation of Uncertainty Routing in Armed Conflict Forecasting ↗
- Why Do AI Agents Break Rules? How Framing, Context, and Social Signals Shape Compliance ↗
Original source text
arXiv:2608.12335v1 Announce Type: new Abstract: Financial question answering over annual reports requires more than retrieving semantically similar passages. It often involves identifying relevant companies and fiscal years, locating standardized filing sections, collecting textual and tabular evidence, and checking answers against the original documents. Existing RAG systems, however, usually flatten long filings into unordered chunks, pay limited attention to the typed structure of financial reports, and use fixed text-table fusion strategies without considering query intent. To address these limitations, we propose , a hierarchical cross-modal retrieval-augmented generation framework for evidence-centric financial QA. HC-RAG organizes filings into a typed financial evidence graph with documents, sections, text units, table units, and metadata nodes. It retrieves evidence through document-section-unit paths, aligns textual and tabular evidence in a shared retrieval space, and routes evidence according to four semantic intents: calculation, trend, fact, and comparison. We further introduce , a benchmark containing 2,327 expert-verified QA pairs from 179 SEC 10-K filings of 87 S&P 500 companies across fiscal years 2022--2024, with labels for intent, difficulty, and structural evidence attributes. Experiments on public financial QA benchmarks and Multi-Doc-2025 show that HC-RAG improves both answer quality and evidence localization, especially in long-document, table-related, and cross-document settings. HC-RAG outperforms RAPTOR by 6.6 F1 points on DocFinQA and GraphRAG by 10.9 F1 points on Multi-Doc-2025. Evidence-level analysis and ablation studies show that the improvements mainly come from more accurate section localization, table grounding, cross-document evidence aggregation, and intent-aware text-table routing.