Classifying Relation via Bidirectional Recurrent Neural Network Based on Local Information

摘要：

　　Relation classification is an important research task in the field of natural language processing (NLP).In this paper,we apply a bidirectional recurrent neural network upon local windows of entities for relation classification.In contrast to previous approaches,only word tokens around entities are taken into consideration in our model.Upon word tokens,a bidirectional recurrent neural network is used to extract local context features of entities.To retain the important features for classification,we propose to use a novel weighted pooling layer upon hidden layers of RNN.Experiments on the SemEval-2010 dataset show that our proposed method achieves competitive results without introducing any external resources.

关键词： Relation extraction Bidirectional recurrent neural network Weighted pooling

作者: Xiaoyun Hou Zhe Zhao Tao Liu Xiaoyong Du

作者单位: School of Information,Renmin University of China,Beijing,China

会议类型: 国际会议

会议名称: International Asia-Pacific Web Conference(第18届国际亚太互联网大会)

会议地点: 苏州

会议语种:英文

页码: 420-430

在线出版日期: 2016-09-23（万方平台首次上网日期，不代表论文的发表时间）