SlideTailor: Personalized Presentation Slide Generation for Scientific Papers

2025-12-23

AI Summary

To address the challenge of significant user preference heterogeneity and insufficient personalization in academic presentation generation, this paper introduces the first conditional slide generation task grounded in implicit preferences—namely, exemplar slide pairs and visual templates. Methodologically, we propose a human-behavior-inspired agent framework that integrates multimodal prompt modeling with example-driven preference learning, and design a Chain-of-Speech mechanism for joint generation of spoken narration and editable, structured slides. Our approach uniquely enables implicit preference distillation and cross-template generalization. Evaluated on the first user-preference-aware benchmark, our method significantly outperforms existing baselines: generated slides better align with users’ stylistic preferences and communicative intent, while the unified speech-slide output natively supports downstream applications such as video-based presentations.

Technology Category

• Knowledge Representation and Reasoning: Preferences
• Humans and AI: Learning Human Values and Preferences
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Automatic presentation slide generation can greatly streamline content creation. However, since preferences of each user may vary, existing under-specified formulations often lead to suboptimal results that fail to align with individual user needs. We introduce a novel task that conditions paper-to-slides generation on user-specified preferences. We propose a human behavior-inspired agentic framework, SlideTailor, that progressively generates editable slides in a user-aligned manner. Instead of requiring users to write their preferences in detailed textual form, our system only asks for a paper-slides example pair and a visual template - natural and easy-to-provide artifacts that implicitly encode rich user preferences across content and visual style. Despite the implicit and unlabeled nature of these inputs, our framework effectively distills and generalizes the preferences to guide customized slide generation. We also introduce a novel chain-of-speech mechanism to align slide content with planned oral narration. Such a design significantly enhances the quality of generated slides and enables downstream applications like video presentations. To support this new task, we construct a benchmark dataset that captures diverse user preferences, with carefully designed interpretable metrics for robust evaluation. Extensive experiments demonstrate the effectiveness of our framework.

Problem

• Generates personalized slides from scientific papers
• Aligns slide content with user preferences and narration
• Uses example pairs and templates to infer user styles

Innovation

• Generates slides using user-provided example pairs and templates
• Employs a chain-of-speech mechanism for narration-aligned content
• Distills implicit preferences from unlabeled inputs for customization