Character-based LSTM-CRF with radical-level features for chinese named entity recognition

Abstract

State-of-the-art systems of Chinese Named Entity Recognition (CNER) require large amounts of hand-crafted features and domainspecific knowledge to achieve high performance. In this paper, we apply a bidirectional LSTM-CRF neural network that utilizes both characterlevel and radical-level representations. We are the first to use characterbased BLSTM-CRF neural architecture for CNER. By contrasting the results of different variants of LSTM blocks, we find the most suitable LSTM block for CNER. We are also the first to investigate Chinese radical-level representations in BLSTM-CRF architecture and get better performance without carefully designed features.We evaluate our system on the third SIGHAN Bakeoff MSRA data set for simplfied CNER task and achieve state-of-the-art performance 90.95% F1.