Position Masking for Improved Layout-Aware Document Understanding
This work addresses the need for more efficient business processes through better NLP for document scans and PDFs, but it is incremental as it builds on existing layout-aware word embeddings.
The paper tackles the problem of improving layout-aware document understanding by introducing a new pre-training task called position masking, which enhances the performance of models like LayoutLM by over 5% on a form understanding task.
Natural language processing for document scans and PDFs has the potential to enormously improve the efficiency of business processes. Layout-aware word embeddings such as LayoutLM have shown promise for classification of and information extraction from such documents. This paper proposes a new pre-training task called that can improve performance of layout-aware word embeddings that incorporate 2-D position embeddings. We compare models pre-trained with only language masking against models pre-trained with both language masking and position masking, and we find that position masking improves performance by over 5% on a form understanding task.