基于全局和局部特征感知网络的关系提取方法

Global and Local Feature-Aware Network for Relation Extraction

关系提取旨在从大量的非结构化数据中提取实体对间的关系。目前,大多数神经网络仅考虑单分支信息流,难以获取足够的语义特征来进行关系提取。针对此问题,该文提出一种基于全局和局部特征感知网络的关系提取方法。该方法首先利用自注意力机制和循环神经网络获取单词的相关性时序特征。其次,构建多分支特征感知卷积神经网络,分别获取全局和局部相关性时序特征,避免全局和局部感知的相互影响。进一步地,对这两种特征进行拼接和筛选,以全面表示句子的重要语义特征。最后,结合Softmax分类器实现关系提取。实验结果表明,该文所提方法优于主流的基于卷积神经网络和循环神经网络的关系提取方法,在标准的SemEval-2010 Task 8和KBP37数据集上F1值分别达到86.1%和64.9%。

Relation extraction aims to identify the relationship between entities from a large amount of unstructured data.This paper proposes a relation extraction method based on global and local feature-aware network.This method first adopts the self-attention mechanism and recurrent neural network to obtain the correlated sequence features of each word.Then,a multi-branch feature-aware convolutional neural network is constructed to obtain global and local features without their mutual interference.Moreover,the obtained two features are concatenated to fully represent the important semantic features of the sentence.Experimental results show that the proposed method performs better than the state-of-the-art methods based on convolutional neural networks and recurrent neural networks,with the F1 of 86.1%and 64.9%on the standard SemEval-2010 Task 8 and KBP37 datasets,respectively.