[论文] DeepSlide: From Artifacts to Presentation Delivery

论文概要

研究领域: NLP 作者: Ming Yang, Zhiwei Zhang, Jiahang Li 发布时间: 2025-05-15 arXiv: 2505.10892

中文摘要

演示文稿是学术交流的主要媒介，然而大多数AI幻灯片生成器只优化产物（视觉上可信的幻灯片组），却忽略了交付过程（节奏、叙事和演讲准备）。我们提出DeepSlide，一种人在回路的多智能体系统，支持完整的演示准备流程，从需求获取、时间预算的叙事规划，到基于证据的幻灯片-脚本生成、注意力增强和排练支持。DeepSlide集成了（i）带节点时间预算的可控逻辑链规划器，（ii）轻量级内容树检索器用于 grounding，（iii）带样式继承的马尔可夫式顺序渲染，以及（iv）沙箱执行与最小修复以确保可渲染性。我们进一步引入双记分牌基准，将静态产物质量与动态交付表现 cleanly 分离。跨越20个领域和多样化受众，DeepSlide在产物质量上匹敌强基线，同时在交付指标上持续取得更大提升，改善叙事流畅度、节奏精度和幻灯片-脚本协同。

原文摘要

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). 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. 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. ...

自动采集于 2026-05-19

#论文 #arXiv #NLP #小凯