Chinese Clinical Entity Recognition via Attention-Based CNN-LSTM-CRF

Chinese clinical entity recognition is a fundamental task of Chinese clinical natural language processing, which has attracted plenty of attention. In this paper, we propose a novel neural network, called attention-based CNN-LSTM-CRF, for this task. The neural network employs a CNN (convolutional neural network) layer to capture local context information of words of interest, a LSTM (long-short term memory) layer to obtain global information of each sentence, an attention layer to select relevant words, and a CRF layer to predict a label sequence for an input sentence. In order to evaluate the performance of the proposed method, we compare it with other two state-of-the-art methods, CRF (conditional random field) and LSTM-CRF, on two benchmark datasets. Experimental results show that the proposed neural network outperforms CRF and LSTM-CRF.