对比两种引用生成方式，为高风险场景选型提供依据。
Generation-Time vs. Post-hoc Citation: A Holistic Evaluation of LLM Attribution
- 分生成时引用和事后引用两种模式，统一评估标准。
- 事后引用覆盖更全、正确率高，生成时引用更精准但漏引多。
- 建议高风险场景优先用事后引用，严苛验证用生成时引用。
可信的大语言模型在医疗、法律、学术和金融等高风险领域需引用可验证的人类来源，微小错误可能带来严重后果。研究人员面临抉择：在生成回答时同步生成引用（生成时引用，G-Cite），或先撰写内容再补充引用（事后引用，P-Cite）。本文提出两种范式，并在四个主流引用数据集上，从零样本到先进检索增强方法进行全面评估，提供基于证据的使用建议。结果表明，在两种范式中，检索是提升引用质量的核心因素。P-Cite 方法在保持较高正确率的同时实现高覆盖率与适中延迟，而 G-Cite 方法以牺牲覆盖率和速度为代价换取更高精度。建议高风险应用采用以检索为中心的 P-Cite 首选策略，仅在严格声明验证等精度敏感场景保留 G-Cite。代码与人工评估结果已公开于 https://anonymous.4open.science/r/Citation_Paradigms-BBB5/
原文摘要 · Abstract (English)
Trustworthy Large Language Models (LLMs) must cite human-verifiable sources in high-stakes domains such as healthcare, law, academia, and finance, where even small errors can have severe consequences. Practitioners and researchers face a choice: let models generate citations during decoding, or let models draft answers first and then attach appropriate citations. To clarify this choice, we introduce two paradigms: Generation-Time Citation (G-Cite), which produces the answer and citations in one pass, and Post-hoc Citation (P-Cite), which adds or verifies citations after drafting. We conduct a comprehensive evaluation from zero-shot to advanced retrieval-augmented methods across four popular attribution datasets and provide evidence-based recommendations that weigh trade-offs across use cases. Our results show a consistent trade-off between coverage and citation correctness, with retrieval as the main driver of attribution quality in both paradigms. P-Cite methods achieve high coverage with competitive correctness and moderate latency, whereas G-Cite methods prioritize precision at the cost of coverage and speed. We recommend a retrieval-centric, P-Cite-first approach for high-stakes applications, reserving G-Cite for precision-critical settings such as strict claim verification. Our codes and human evaluation results are available at https://anonymous.4open.science/r/Citation_Paradigms-BBB5/
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。