Improved Relation Classification by Deep Recurrent Neural Networks with Data Augmentation

TL;DR

This paper introduces deep recurrent neural networks with data augmentation for relation classification, significantly improving performance over previous shallow models by exploring multiple abstraction levels.

Contribution

The paper presents a novel deep RNN architecture and a data augmentation method leveraging relation directionality, advancing relation classification accuracy.

Findings

Achieved an F1-score of 86.1% on SemEval-2010 Task 8.

Outperformed previous state-of-the-art results.

Demonstrated effectiveness of deep RNNs with data augmentation.

Abstract

Nowadays, neural networks play an important role in the task of relation classification. By designing different neural architectures, researchers have improved the performance to a large extent in comparison with traditional methods. However, existing neural networks for relation classification are usually of shallow architectures (e.g., one-layer convolutional neural networks or recurrent networks). They may fail to explore the potential representation space in different abstraction levels. In this paper, we propose deep recurrent neural networks (DRNNs) for relation classification to tackle this challenge. Further, we propose a data augmentation method by leveraging the directionality of relations. We evaluated our DRNNs on the SemEval-2010 Task~8, and achieve an F1-score of 86.1%, outperforming previous state-of-the-art recorded results.