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

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1405–1418 June 6–11, 2021. ©2021 Association for Computational Linguistics 1405 D2S: Document-to-Slide Generation Via Query-Based Text Summarization Edward Sun∗ University of Michigan Michigan, USA edwsun@umich.edu Yufang Hou IBM Research Europe Dublin, Ireland yhou@ie.ibm.com Dakuo Wang IBM Research Cambridge, USA dakuo.wang@ibm.com Yunfeng Zhang IBM Research Yorktown, USA zhangyun@us.ibm.com Nancy X.R. Wang IBM Research Almaden, USA nancywang1991@gmail.com 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 au- tomate the document-to-slides generation pro- cess and all face a critical challenge: no pub- licly available dataset for training and bench- marking. In this work, we first contribute a new dataset, SciDuet, consisting of pairs of pa- pers and their corresponding slides decks from recent years’ NLP and ML conferences (e.g., ACL). Secondly, we present D2S, a novel sys- tem 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 con- text into bullet points with long-form question answering. Our evaluation suggests that long- form QA outperforms state-of-the-art summa- rization baselines on both automated ROUGE metrics and qualitative human evaluation. 1 Introduction From business to education to research, presenta- tions are everywhere as they are visually effective in summarizing and explaining bodies of work to the audience (Bartsch and Cobern, 2003; Wang, 2016; Piorkowski et al., 2021). However, it is te- dious and time-consuming to manually create pre- sentation slides (Franco et al., 2016). Researchers have proposed various methods to automatically generate presentations from source documents. For example, Winters and Mathewson (2019) suggest heuristic rule-based mechanisms to extract document contents and use those as the generated-slide’s content. PPSGen (Hu and Wan, 2014) leverages machine learning (ML) approaches ∗Work done during internship at IBM Research. Figure 1: An example slide in SciDuet. TOP is author’s original slide; BOTTOM is from our system D2S. to learn a sentence’s importance in the document, and extract important sentences as slide’s content. These existing research works have yielded promising progress towards the goal of automated slide generation, but they also face two com- mon limitations: 1) these works primarily rely on extractive-based mechanisms, thus the generated content is merely an aggregation of raw sentences from the document, whereas in real-world slides, the presenter frequently uses abstractive summa- rization; 2) these works assume the presentation slide’s title has a one-to-one match to the docu- ment’s subtitles or section headlines, whereas the presenter in reality often uses new slide titles and creates multiple slides under the same title (e.g., 1406 the slides with a Cont. / Continue on it). In our work, we aim to tackle both the limi- tations. To achieve this goal, we consider the document-to-slides generation task as a Query- Based Single-Document Text Summarization (QSS) task. Our approach leverages recent research devel- …

Content truncated to 500 words