D2S: Document-to-Slide Generation Via Query-Based Text Summarization

TL;DR

This paper introduces SciDuet, a new dataset for document-to-slide generation, and D2S, a system that creates presentation slides from papers using retrieval and long-form question answering, improving over existing methods.

Contribution

The paper provides a new dataset SciDuet and a novel two-step system D2S for automated document-to-slide generation, advancing research in this area.

Findings

Long-form QA outperforms summarization baselines in ROUGE scores.

D2S generates more engaging and relevant slides.

The dataset enables benchmarking for future research.

Abstract

Presentations are critical for communication in all areas of our lives, yet the creation of slide decks is often tedious and time-consuming. There has been limited research aiming to automate the document-to-slides generation process and all face a critical challenge: no publicly available dataset for training and benchmarking. In this work, we first contribute a new dataset, SciDuet, consisting of pairs of papers and their corresponding slides decks from recent years' NLP and ML conferences (e.g., ACL). Secondly, we present D2S, a novel system that tackles the document-to-slides task with a two-step approach: 1) Use slide titles to retrieve relevant and engaging text, figures, and tables; 2) Summarize the retrieved context into bullet points with long-form question answering. Our evaluation suggests that long-form QA outperforms state-of-the-art summarization baselines on both automated ROUGE metrics and qualitative human evaluation.