Chinese Word Segmentation Based on Bi-GRU Integrating Dictionary Information
Rong Xiang, Shiqun Yin · 2020
Chinese word segmentation (CWS) is an important and essential pre-processing step for Chinese language processing tasks. To date, various models based on deep neural networks have been extensively applied in CWS. Most of them learn from large scale labeled data. However, these models typically lack the capability of processing rare words and OOV words. In this paper, we use character embedding and bigram embedding as the inputs of Bi-GRU model and construct a feature vector to capture dictionary information of characters. we use the multi-head attention mechanism to get the weight of different features which is the inputs of a parallel Bi-GRU network. To evaluate the performance of the proposed model, we conducted experiments on PKU and MSR datasets. The experimental results of datasets show that our model achieves the-state-of-art performance compared to several baselines.