HIT: Nested Named Entity Recognition via Head-Tail Pair and Token Interaction

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 6027–6036, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 6027 HIT: Nested Named Entity Recognition via Head-Tail Pair and Token Interaction Yu Wang1, Yun Li1∗, Hanghang Tong2, Ziye Zhu1 1Jiangsu Key Laboratory of Big Data Security and Intelligent Processing Nanjing University of Posts and Telecommunications, Nanjing, China 2Department of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL {2017070114, liyun, 2016070251}@njupt.edu.cn, htong@illinois.edu Abstract Named Entity Recognition (NER) is a fun- damental task in natural language processing. In order to identify entities with nested struc- ture, many sophisticated methods have been recently developed based on either the tra- ditional sequence labeling approaches or di- rected hypergraph structures. Despite being successful, these methods often fall short in striking a good balance between the expres- sion power for nested structure and the model complexity. To address this issue, we present a novel nested NER model named HIT. Our pro- posed HIT model leverages two key properties pertaining to the (nested) named entity, includ- ing (1) explicit boundary tokens and (2) tight internal connection between tokens within the boundary. Specifically, we design (1) Head- Tail Detector based on the multi-head self- attention mechanism and bi-affine classifier to detect boundary tokens, and (2) Token Interac- tion Tagger based on traditional sequence la- beling approaches to characterize the internal token connection within the boundary. Exper- iments on three public NER datasets demon- strate that the proposed HIT achieves state-of- the-art performance. 1 Introduction Named Entity Recognition (NER) is a fundamen- tal task in natural language processing due to the fact that the named entities often convey the key information of the text (Lample et al., 2016). It is common in many practical scenarios that the named entities have a nested structure (Finkel and Man- ning, 2009; Silla and Freitas, 2011). That is, an en- tity could contain other entities or be a part of other entities. As shown in Figure 1, the entity “the west- ern Canadian province of British Columbia” in the first example contains two inner entities, i.e., “west- ern Canadian” and “British Columbia”. Traditional methods often treat the NER task as a sequence Figure 1: Examples of the named entity. The first ex- ample is a sentence with nested named entities, and the second one is a sentence only with flat named entities. labeling problem (Lin et al., 2018) and thus are primarily designed to recognize flat entities in the input sentences (as shown in the second example in Figure 1). Due to the nature of the nested en- tity, a token might belong to different entities. It is difficult to represent such nested structures using a single label accurately. Therefore, the performance of traditional NER methods will dramatically suf- fer when recognizing nested entities (Katiyar and Cardie, 2018). In recent years, more sophisticated methods have been developed for the nested NER task, which are grouped into two categories, including sequence- based method and hypergraph-based method. The sequence-based methods (Sohrab …

Content truncated to 500 words