DeepSlide: From Artifacts to Presentation Delivery

DeepSlide：从演示文档到演讲交付

Presentations are a primary medium for scholarly communication, yet most AI slide generators optimize the artifact (a visually plausible deck) while under-optimizing the delivery process (pacing, narrative, and presentation preparation). 演示文稿是学术交流的主要媒介，然而大多数人工智能幻灯片生成器仅优化了演示文档（即视觉上合理的幻灯片组），却忽视了对交付过程（如节奏控制、叙事逻辑和演讲准备）的优化。

We present DeepSlide, a human-in-the-loop multi-agent system that supports preparing the full presentation process, from requirement elicitation and time-budgeted narrative planning, to evidence-grounded slide—script generation, attention augmentation, and rehearsal support. 我们提出了 DeepSlide，这是一个“人在回路”（human-in-the-loop）的多智能体系统，旨在支持整个演示准备过程，涵盖需求获取、基于时间预算的叙事规划，以及基于证据的幻灯片与讲稿生成、注意力增强和排练支持。

DeepSlide integrates (i) a controllable logical-chain planner with per-node time budgets, (ii) a lightweight content-tree retriever for grounding, (iii) Markov-style sequential rendering with style inheritance, and (iv) sandboxed execution with minimal repair to ensure renderability. DeepSlide 集成了以下核心功能：(i) 带有节点级时间预算的可控逻辑链规划器；(ii) 用于内容溯源的轻量级内容树检索器；(iii) 具备样式继承功能的马尔可夫式序列渲染；以及 (iv) 带有最小化修复机制的沙盒执行环境，以确保渲染的可行性。

We further introduce a dual-scoreboard benchmark that cleanly separates static artifact quality from dynamic delivery excellence. 此外，我们引入了一个双重计分基准测试，能够清晰地将静态演示文档质量与动态演讲交付表现区分开来。

Across 20 domains and diverse audience profiles, DeepSlide matches strong baselines on artifact quality while consistently achieving larger gains on delivery metrics, improving narrative flow, pacing precision, and slide—script synergy with clearer attention guidance. 在涵盖 20 个领域和不同受众画像的测试中，DeepSlide 在演示文档质量方面与强基准模型持平，同时在交付指标上持续取得显著提升，通过更清晰的注意力引导，改善了叙事流畅度、节奏精准度以及幻灯片与讲稿之间的协同效应。