MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding
This addresses the challenge of processing interactive digital documents for applications like web content analysis, though it is incremental as it builds on existing multimodal pre-training approaches.
The authors tackled the problem of understanding digital documents with dynamic layouts, such as HTML/XML, by proposing MarkupLM, a pre-training method that jointly learns text and markup information. The model significantly outperformed existing baselines on several document understanding tasks.
Multimodal pre-training with text, layout, and image has made significant progress for Visually Rich Document Understanding (VRDU), especially the fixed-layout documents such as scanned document images. While, there are still a large number of digital documents where the layout information is not fixed and needs to be interactively and dynamically rendered for visualization, making existing layout-based pre-training approaches not easy to apply. In this paper, we propose MarkupLM for document understanding tasks with markup languages as the backbone, such as HTML/XML-based documents, where text and markup information is jointly pre-trained. Experiment results show that the pre-trained MarkupLM significantly outperforms the existing strong baseline models on several document understanding tasks. The pre-trained model and code will be publicly available at https://aka.ms/markuplm.