Discontinuous Named Entity Recognition as Maximal Clique Discovery

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 764–774 August 1–6, 2021. ©2021 Association for Computational Linguistics 764 Discontinuous Named Entity Recognition as Maximal Clique Discovery Yucheng Wang ∗, Bowen Yu ∗, Hongsong Zhu Tingwen Liu †, Nan Yu, Limin Sun Institute of Information Engineering, Chinese Academy of Sciences School of Cyber Security, University of Chinese Academy of Sciences {wangyucheng,yubowen,zhuhongsong}@iie.ac.cn {liutingwen,yunan,sunlimin}@iie.ac.cn Abstract Named entity recognition (NER) remains chal- lenging when entity mentions can be discontin- uous. Existing methods break the recognition process into several sequential steps. In train- ing, they predict conditioned on the golden intermediate results, while at inference rely- ing on the model output of the previous steps, which introduces exposure bias. To solve this problem, we first construct a segment graph for each sentence, in which each node denotes a segment (a continuous entity on its own, or a part of discontinuous entities), and an edge links two nodes that belong to the same en- tity. The nodes and edges can be generated respectively in one stage with a grid tagging scheme and learned jointly using a novel ar- chitecture named Mac. Then discontinuous NER can be reformulated as a non-parametric process of discovering maximal cliques in the graph and concatenating the spans in each clique. Experiments on three benchmarks show that our method outperforms the state-of- the-art (SOTA) results, with up to 3.5 percent- age points improvement on F1, and achieves 5x speedup over the SOTA model.1 1 Introduction Named Entity Recognition (NER) is the task of detecting mentions of real-world entities from text and classifying them into predefined types. NER benefits many natural language processing applica- tions (e.g., information retrieval (Berger and Laf- ferty, 2017), relation extraction (Yu et al., 2019), and question answering (Khalid et al., 2008)). NER methods have been extensively investigated and researchers have proposed effective ones. Most prior approaches (Huang et al., 2015; Chiu and ∗ The two authors contribute equally. † Corresponding author. 1The source code is available at https://github. com/131250208/InfExtraction productive cough with white or bloody sputum E1 E1 E1 E2 E2 E2 Figure 1: An example involving discontinuous men- tions. Entities are highlighted with colored underlines. Nichols, 2016; Gridach, 2017; Zhang and Yang, 2018; Gui et al., 2019; Xue et al., 2020) cast this task as a sequence labeling problem where each token is assigned a label that represents its entity type. Their underlying assumption is that an entity mention should be a short span of text (Muis and Lu, 2016), and should not overlap with each other. While such assumption is valid for most cases, it does not always hold, especially in clinical cor- pus (Pradhan et al., 2015). For example, Figure 1 shows two discontiguous entity mentions with over- lapping segments. Thus, there is a need to move beyond continuous entities and devise methods to extract discontinuous ones. Towards this goal, current state-of-the-art (SOTA) models can be categorized into two classes: …

Content truncated to 500 words