Chinese Named Entity Recognition via Adaptive Multi-pass Memory Network with Hierarchical Tagging Mechanism

Chinese Named Entity Recognition via Adaptive Multi-pass Memory Network with Hierarchical Tagging Mechanism Pengfei Cao1,2, Yubo Chen1,2, Kang Liu1,2 and Jun Zhao1,2 1 National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China 2 School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, 100049, China {pengfei.cao, yubo.chen, kliu, jzhao}@nlpr.ia.ac.cn Abstract Named entity recognition (NER) aims to identify text spans that mention named entities and classify them into pre-defined categories. For Chinese NER task, most of the existing methods are character-based sequence labeling models and achieve great success. However, these methods usually ignore lexical knowledge, which leads to false prediction of entity boundaries. Moreover, these methods have difficulties in capturing tag dependencies. In this paper, we propose an Adaptive Multi-pass Memory Network with Hierarchical Tagging Mechanism (AMMNHT) to address all above problems. Specifically, to reduce the errors of predicting entity boundaries, we propose an adaptive multi-pass memory network to exploit lexical knowledge. In addition, we propose a hierarchical tagging layer to learn tag dependencies. Experimental results on three widely used Chinese NER datasets demonstrate that our proposed model outperforms other state- of-the-art methods. 1 Introduction The task of named entity recognition (NER) is to recognize the named entities from a plain text and classify them into pre-defined types. NER is a fundamental and preliminary task in natural language processing (NLP) area and is beneficial for many downstream NLP tasks such as relation extraction (Bunescu and Mooney, 2005), event extraction (Chen et al., 2015) and question answering (Yahya et al., 2013). In recent years, numerous methods have been carefully studied for NER task, including Conditional Random Fields (CRFs) (Lafferty et al., 2001) and Support Vector Machines (SVMs) (Isozaki and Kazawa, 2002). Currently, with the development of deep learning methods, neural networks have been introduced for the NER task. In particular, sequence labeling neural network models have achieved state-of-the-art performance (Lample et al., 2016; Zhang and Yang, 2018). Though sequence labeling neural network methods have achieved great success for Chinese NER task, some challenging issues still have not been well addressed. One significant drawback is that previous methods usually fail to correctly predict entity boundaries. To conduct a quantitative analysis, we perform a BiLSTM+CRF model proposed by Huang et al. (2015), which is the most representative Chi- nese NER sequence labeling system, on WeiboNER dataset (Peng and Dredze, 2015; He and Sun, 2016), OntoNotes 4 dataset (Weischedel et al., 2011) and MSRA dataset (Levow, 2006). The F1 scores are 55.84%, 63.17% and 89.13%, respectively. We do a further analysis and find that the errors of predicting entity boundaries are particularly serious. The average proportion of predicting entity boundaries errors is 82% on these three datasets. For example, the character-based BiLSTM+CRF model fails to predict the entity boundaries of “北海道 (Hokkaido)” in Figure 1. To reduce the errors of predicting entity boundaries, some works (Peng and Dredze, 2016; Cao et al., 2018) try to jointly perform Chinese NER with Chinese word segmentation (CWS) for using word boundaries …

Content truncated to 500 words