# MHMS: Multimodal Hierarchical Multimedia Summarization

**Authors:** Jielin Qiu, Jiacheng Zhu, Mengdi Xu, Franck Dernoncourt, Trung Bui, Zhaowen Wang, Bo Li, Ding Zhao, Hailin Jin  
**arXiv:** [2204.03734](https://arxiv.org/abs/2204.03734)  
**Date:** 2022-04-07  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2204.03734v1/

## Key points

- The paper introduces a multimodal hierarchical multimedia summarization (MHMS) framework
- The framework combines visual and language domains to generate video and textual summaries
- It consists of video and textual segmentation and summarization modules
- Cross-domain alignment objective with optimal transport distance is utilized to generate representative keyframes and textual summaries
- Evaluations on three multimodal datasets demonstrate the effectiveness of the MHMS method in producing high-quality multimodal summaries
- Authors: Jielin Qiu, Jiacheng Zhu, Mengdi Xu, Franck Dernoncourt, Trung Bui, Zhaowen Wang, Bo Li, Ding Zhao, Hailin Jin

## Summary

The paper introduces a multimodal hierarchical multimedia summarization (MHMS) framework that combines visual and language domains to generate video and textual summaries. The framework consists of video and textual segmentation and summarization modules which utilize a cross-domain alignment objective with optimal transport distance to generate representative keyframes and textual summaries. Evaluations on three multimodal datasets demonstrate the effectiveness of the MHMS method in producing high-quality multimodal summaries. The authors of the paper are Jielin Qiu, Jiacheng Zhu, Mengdi Xu, Franck Dernoncourt, Trung Bui, Zhaowen Wang, Bo Li, Ding Zhao, and Hailin Jin.