Bag of Tricks for Chinese Named Entity Recognition

Named entity recognition (NER) is an important and challenging task in natural language processing. In this paper, we investigate thoroughly about the advances of Chinese NER in recent years. We explore the validity of a wide range of approaches in the literature of NLP that may benefit NER. We further employ the effective ones, such as data augmentation, adversarial learning, cross-sentence context and cost-sensitive learning to improve the performance of our BERT-based backbone model. Empirical results show that our model with this bag of tricks outperforms previous state-of-the-art on Weibo and achieves competitive performance on MSRA.