SlideTailor: Personalized Presentation Slide Generation for Scientific Papers
Wenzheng Zeng, Mingyu Ouyang, Langyuan Cui, Hwee Tou Ng
2025-12-29
Summary
This paper introduces a new way to automatically create presentation slides from research papers, but with a focus on making those slides match *your* specific style and preferences.
What's the problem?
Currently, automatic slide generators don't do a great job of creating slides that people actually like. They often create slides that are okay, but not tailored to how a specific person wants their presentations to look or what information they emphasize. Asking users to write out detailed instructions for the slides is also a pain.
What's the solution?
The researchers developed a system called SlideTailor. Instead of making you write out preferences, you just show it an example of a paper and the slides someone made from it that you *do* like, along with a visual template. SlideTailor learns your preferences from these examples and uses that to create new slides. It also includes a feature to make sure the slide content matches what you'd say when presenting, making for a more cohesive presentation. They even created a new dataset to test their system.
Why it matters?
This work is important because it makes automatic slide generation much more useful. By learning from examples instead of requiring detailed instructions, it's easier for anyone to create presentations that truly reflect their individual style and needs, and it can even help create scripts for video presentations.
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.