Reading Order Matters: Information Extraction from Visually-rich Documents by Token Path Prediction

17 Oct 2023  ·  Chong Zhang, Ya Guo, Yi Tu, Huan Chen, Jinyang Tang, Huijia Zhu, Qi Zhang, Tao Gui ·

Recent advances in multimodal pre-trained models have significantly improved information extraction from visually-rich documents (VrDs), in which named entity recognition (NER) is treated as a sequence-labeling task of predicting the BIO entity tags for tokens, following the typical setting of NLP. However, BIO-tagging scheme relies on the correct order of model inputs, which is not guaranteed in real-world NER on scanned VrDs where text are recognized and arranged by OCR systems. Such reading order issue hinders the accurate marking of entities by BIO-tagging scheme, making it impossible for sequence-labeling methods to predict correct named entities. To address the reading order issue, we introduce Token Path Prediction (TPP), a simple prediction head to predict entity mentions as token sequences within documents. Alternative to token classification, TPP models the document layout as a complete directed graph of tokens, and predicts token paths within the graph as entities. For better evaluation of VrD-NER systems, we also propose two revised benchmark datasets of NER on scanned documents which can reflect real-world scenarios. Experiment results demonstrate the effectiveness of our method, and suggest its potential to be a universal solution to various information extraction tasks on documents.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Named Entity Recognition (NER) CORD-r TPP (LayoutMask) F1 89.34 # 2
Named Entity Recognition (NER) CORD-r TPP (LayoutLMv3) F1 91.85 # 1
Relation Extraction FUNSD TPP (LayoutMask) F1 79.20 # 4
Entity Linking FUNSD TPP (LayoutMask) F1 79.20 # 1
Named Entity Recognition (NER) FUNSD-r TPP (LayoutLMv3) F1 80.40 # 1
Named Entity Recognition (NER) FUNSD-r TPP (LayoutMask) F1 78.19 # 3
Reading Order Detection ReadingBank TPP (LayoutMask) Average Page-level BLEU 98.16 # 2
Average Relative Distance (ARD) 0.37 # 1

Methods


No methods listed for this paper. Add relevant methods here