An Attention-Based BILSTM-CRF for Chinese Named Entity Recognition
Qing Zhong, Yan Tang · 2020
Named entity recognition (NER) is a very basic task in natural language processing (NLP). Compared with English, the task of Chinese NER faces more challenges. One of the reasons is the blurring of Chinese entity boundaries, which is closely related to the segmentation results. Previous research on this task can be roughly divided into two categories, word-based methods and character-based methods, but both methods have their own shortcomings. In this article, we combine character information with word information, and introduce the attention mechanism into a bidirectional long short-term memory network-conditional random field (BILSTM-CRF) model. First, we utilizes a bidirectional long short-term memory network to obtain more complete contextual information. Then the model utilizes the attention mechanism to learn more about the context grammar of Chinese named entities, and finally, our model uses the conditional random field model to obtain the predicted label sequence. Experimental results T prove that our method can effectively improve the effectiveness of named entity recognition.