Merge and Label: A Novel Neural Network Architecture for Nested NER

Abstract

Named entity recognition (NER) is one of the best studied tasks in natural language processing. However, most approaches are not capable of handling nested structures which are common in many applications. In this paper we introduce a novel neural network architecture that first merges tokens and/or entities into entities forming nested structures, and then labels each of them independently. Unlike previous work, our merge and label approach predicts real-valued instead of discrete segmentation structures, which allow it to combine word and nested entity embeddings while maintaining differentiability. We evaluate our approach using the ACE 2005 Corpus, where it achieves state-of-the-art F1 of 74.6, further improved with contextual embeddings (BERT) to 82.4, an overall improvement of close to 8 F1 points over previous approaches trained on the same data. Additionally we compare it against BiLSTM-CRFs, the dominant approach for flat NER structures, demonstrating that its ability to predict nested structures does not impact performance in simpler cases. 1

Nested Named Entity Recognition via Second-best Sequence Learning and Decoding

“…However, they also use the maximal-length hyperparameter to reduce time complexity. Fisher and Vlachos (2019) proposed a novel neural network architecture that merges tokens or entities into entities forming nested structures and then labels each of them. Their architecture, however, needs the maximal nesting level hyperparameter.…”

A Boundary Assembling Method for Nested Biomedical Named Entity Recognition

“…Lin et al [34] proposed a sequence-to-nugget architecture that uses a head-driven phrase structure for nested NE recognition. In Table 7, the BERT is used in Xia et al [55], Fisher et al [56], Shibuya et al [57], StrakovÂt'a et al [59] and Jue et al [84]. Compared with them, our model achieves state-of-theart performance in the task of nested NE recognition.…”

Hierarchical Region Learning for Nested Named Entity Recognition

“…To investigate the effectiveness and efficiency of our proposed method, we conduct comprehensive experiments on three benchmark NER datasets. (Lin et al, 2019)[POS] 76.2 73.6 74.9 75.8 73.9 74.8 M&L (Fisher and Vlachos, 2019) 75.1 74.1 74.6 ---Bound. Aware (Zheng et al, 2019) ---75.9 73.6 74.7 BENSC (Tan et al, 2020)[POS] 77.1 74.2 75.6 78.9 72.7 75.7 Our Model(LSTM) [POS] 78.5 74.6 76.5 77.4 73.9 75.6 datasets without nested entities are called as flat NER datasets.…”