Improved Relation Classification by Deep Recurrent Neural Networks
with Data Augmentation
Yan Xu,
1,∗,‡
Ran Jia,
1,∗
Lili Mou,
1
Ge Li,
1,†
Yunchuan Chen,
2
Yangyang Lu,
1
Zhi Jin
1,†
1
Key Laboratory of High Confidence Software Technologies (Peking University),
Ministry of Education, China; Institute of Software, Peking University
{xuyan14,lige,luyy11,zhijin}@sei.pku.edu.cn
{jiaran1994,doublepower.mou}gmail.com
2
University of Chinese Academy of Sciences chenyunchuan11@mails.ucas.ac.cn
Abstract
Nowadays, neural networks play an important role in the task of relation classification. By de-
signing different neural architectures, researchers have improved the performance to a large ex-
tent 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 rela-
tion 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 F
1
-score of 86.1%, outperforming previous state-of-the-art recorded results.
1
1 Introduction
Classifying relations between two entities in a given context is an important task in natural language pro-
cessing (NLP). Take the following sentence as an example: “Jewelry and other smaller [valuables]
e
1
were
locked in a [safe]
e
2
or a closet with a deadbolt.” The marked entities valuables and safe are of relation
Content-Container(e
1
, e
2
). Relation classification plays a key role in various NLP applications,
and has become a hot research topic in recent years.
Nowadays, neural network-based approaches have made significant improvement in relation classifi-
cation, compared with traditional methods based on either human-designed features (Kambhatla, 2004;
Hendrickx et al., 2009) or kernels (Bunescu and Mooney, 2005; Plank and Moschitti, 2013). For exam-
ple, Zeng et al. (2014) and Xu et al. (2015a) utilize convolutional neural networks (CNNs) for relation
classification. Xu et al. (2015b) apply long short term memory (LSTM)-based recurrent neural networks
(RNNs) along the shortest dependency path. Nguyen and Grishman (2015) build ensembles of gated
recurrent unit (GRU)-based RNNs and CNNs.
We have noticed that these neural models are typically designed in shallow architectures, e.g., one layer
of CNN or RNN, whereas evidence in the deep learning community suggests that deep architectures are
more capable of information integration and abstraction (Graves et al., 2013; Hermans and Schrauwen,
2013; Irsoy and Cardie, 2014). A natural question is then whether such deep architectures are beneficial
to the relation classification task.
In this paper, we propose the deep recurrent neural networks (DRNNs) to classify relations. The
deep RNNs can explore the representation space in different levels of abstraction and granularity. By
visualizing how RNN units are related to the ultimate classification, we demonstrate that different layers
indeed learn different representations: low-level layers enable sufficient information mix, while high-
level layers are more capable of precisely locating the information relevant to the target relation between
∗
Equal contribution.
†
Corresponding authors.
‡
Yan Xu is currently a research scientist at Inveno Co., Ltd. .
.
1
Code released on https://sites.google.com/site/drnnre/
This work is licenced under a Creative Commons Attribution 4.0 International License. License details: http://
creativecommons.org/licenses/by/4.0/
Accepted by COLING-2016
arXiv:1601.03651v2 [cs.CL] 13 Oct 2016