A hybrid method based on semi-supervised learning for relation extraction in Chinese EMRs

Table 1 Classical relation extraction methods

Classes	Principle	Classic Methods
Manual	Rule-based methods	Dependency parse trees[7]
Machine learning	Based on kernel functions	Convolution kernel[6]
Machine learning	Based on feature vectors	SVM[4],CRF[5]
Deep learning	Based on convolutional neural networks	CNN[8, 9],PCNN[10],ResNet[2]
	Based on recurrent neural network	RNN[11],LSTM[12],GRU[13]
	Graph-based neural networks	GNN[14]

ISSN: 1472-6947