Fusing Multimodal Signals on Hyper-complex Space for Extreme Abstractive Text Summarization (TL;DR) of Scientific Contents

TL;DR

This paper introduces a novel multimodal dataset and a hyper-complex Transformer model for extreme abstractive summarization of scientific content, leveraging videos, audio, and text to generate concise summaries.

Contribution

The paper presents the first dataset for multimodal extreme abstractive summarization and a novel hyper-complex Transformer model that effectively captures modality interactions in a geometric space.

Findings

mTLDRgen outperforms 20 baselines on Rouge scores

Generated summaries are fluent and source-congruent

Model effectively captures multimodal interactions

Abstract

The realm of scientific text summarization has experienced remarkable progress due to the availability of annotated brief summaries and ample data. However, the utilization of multiple input modalities, such as videos and audio, has yet to be thoroughly explored. At present, scientific multimodal-input-based text summarization systems tend to employ longer target summaries like abstracts, leading to an underwhelming performance in the task of text summarization. In this paper, we deal with a novel task of extreme abstractive text summarization (aka TL;DR generation) by leveraging multiple input modalities. To this end, we introduce mTLDR, a first-of-its-kind dataset for the aforementioned task, comprising videos, audio, and text, along with both author-composed summaries and expert-annotated summaries. The mTLDR dataset accompanies a total of 4,182 instances collected from various academic conference proceedings, such as ICLR, ACL, and CVPR. Subsequently, we present mTLDRgen, an encoder-decoder-based model that employs a novel dual-fused hyper-complex Transformer combined with a Wasserstein Riemannian Encoder Transformer, to dexterously capture the intricacies between different modalities in a hyper-complex latent geometric space. The hyper-complex Transformer captures the intrinsic properties between the modalities, while the Wasserstein Riemannian Encoder Transformer captures the latent structure of the modalities in the latent space geometry, thereby enabling the model to produce diverse sentences. mTLDRgen outperforms 20 baselines on mTLDR as well as another non-scientific dataset (How2) across three Rouge-based evaluation measures. Furthermore, based on the qualitative metrics, BERTScore and FEQA, and human evaluations, we demonstrate that the summaries generated by mTLDRgen are fluent and congruent to the original source material.