Narrative-Driven Paper-to-Slide Generation via ArcDeck

arXiv cs.AI / 4/15/2026

📰 NewsSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

ArcDeck is a multi-agent framework for generating slide decks from academic papers by treating the task as structured narrative reconstruction rather than plain text summarization.
The method parses the paper into a discourse tree and builds a global “commitment document” to preserve the source’s high-level intent and logical flow.
It uses iterative, role-specialized agents that critique and revise the presentation outline before producing the final slide layouts and designs.
The paper also introduces ArcBench, a newly curated benchmark dataset of academic paper-to-slide pairs to evaluate the approach.
Experiments indicate that explicit discourse modeling plus coordinated agent roles improves narrative flow and logical coherence in the generated presentations.

Abstract

We introduce ArcDeck, a multi-agent framework that formulates paper-to-slide generation as a structured narrative reconstruction task. Unlike existing methods that directly summarize raw text into slides, ArcDeck explicitly models the source paper's logical flow. It first parses the input to construct a discourse tree and establish a global commitment document, ensuring the high-level intent is preserved. These structural priors then guide an iterative multi-agent refinement process, where specialized agents iteratively critique and revise the presentation outline before rendering the final visual layouts and designs. To evaluate our approach, we also introduce ArcBench, a newly curated benchmark of academic paper-slide pairs. Experimental results demonstrate that explicit discourse modeling, combined with role-specific agent coordination, significantly improves the narrative flow and logical coherence of the generated presentations.