Combined Attention Mechanism for Named Entity Recognition in Chinese Electronic Medical Records

Abstract

In order to improve the performance of named entity recognition (NER) in Chinese electronic medical records (EMRs), this paper proposed a deep neural model (BiLSTM-Att-CRF model) that combined bidirectional long-short time memory network(BiLSTM) with an attention mechanism, in terms of simultaneously identifying 5 types of clinical entities from the China Conference on Knowledge Graph and Semantic Computing (CCKS) 2018 Chinese EMRs corpus, the BiLSTM-Att-CRF model finally achieved better performance than other widely-used models (F-score of 85.79% without additional features and the F-Score of 86.35% with the part of speech and dictionary features).