CUTIE: Learning to Understand Documents with Convolutional Universal Text Information Extractor

TL;DR

This paper introduces CUTIE, a convolutional neural network model that effectively extracts structured information from documents by leveraging both semantic and spatial text features, outperforming previous NER-based methods.

Contribution

The paper proposes a novel CNN-based model, CUTIE, that captures both semantic and spatial information from document texts, achieving state-of-the-art results without pre-training or post-processing.

Findings

Achieves superior accuracy and speed over NER-based methods.

Performs well with limited training data.

Does not require pre-training or post-processing.

Abstract

Extracting key information from documents, such as receipts or invoices, and preserving the interested texts to structured data is crucial in the document-intensive streamline processes of office automation in areas that includes but not limited to accounting, financial, and taxation areas. To avoid designing expert rules for each specific type of document, some published works attempt to tackle the problem by learning a model to explore the semantic context in text sequences based on the Named Entity Recognition (NER) method in the NLP field. In this paper, we propose to harness the effective information from both semantic meaning and spatial distribution of texts in documents. Specifically, our proposed model, Convolutional Universal Text Information Extractor (CUTIE), applies convolutional neural networks on gridded texts where texts are embedded as features with semantical connotations. We further explore the effect of employing different structures of convolutional neural network and propose a fast and portable structure. We demonstrate the effectiveness of the proposed method on a dataset with up to $4,484$ labelled receipts, without any pre-training or post-processing, achieving state of the art performance that is much better than the NER based methods in terms of either speed and accuracy. Experimental results also demonstrate that the proposed CUTIE model being able to achieve good performance with a much smaller amount of training data.