Exploit Feature and Relation Hierarchy for Relation Extraction

Abstract

Existing methods in relation extraction have leveraged the lexical features in the word sequence and the syntactic features in the parse tree. Though effective, the lexical features extracted from the successive word sequence may introduce some noise that has little or no meaningful content. Meanwhile, the syntactic features are usually encoded via graph convolutional networks which have restricted receptive field. In addition, the relation between lexical and syntactic features in the representation space has been largely neglected. To address the above limitations, we propose a multi-scale representation and metric learning framework to exploit the feature and relation hierarchy for RE tasks. Methodologically, we <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">build a lexical and syntactic feature and relation hierarchy</i> in text data. Technically, we first develop <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a multi-scale convolutional neural network</i> to aggregate the non-successive lexical patterns in the word sequence. We also design <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a multi-scale graph convolutional network</i> to increase the receptive field via the coarsened syntactic graph. Moreover, we present <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a multi-scale metric learning</i> paradigm to exploit both the feature-level relation between lexical and syntactic features and the sample-level relation between instances with the same or different classes. Extensive experiments on three public datasets for two RE tasks prove that our model achieves a new state-of-the-art performance.