Named Entity Recognition as Dependency Parsing

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6470–6476 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 6470 Named Entity Recognition as Dependency Parsing Juntao Yu Queen Mary University London, UK juntao.yu@qmul.ac.uk Bernd Bohnet Google Research Netherlands bohnetbd@google.com Massimo Poesio Queen Mary University London, UK m.poesio@qmul.ac.uk Abstract Named Entity Recognition (NER) is a funda- mental task in Natural Language Processing, concerned with identifying spans of text ex- pressing references to entities. NER research is often focused on flat entities only (flat NER), ignoring the fact that entity references can be nested, as in [Bank of [China]] (Finkel and Manning, 2009). In this paper, we use ideas from graph-based dependency parsing to pro- vide our model a global view on the input via a biaffine model (Dozat and Manning, 2017). The biaffine model scores pairs of start and end tokens in a sentence which we use to explore all spans, so that the model is able to predict named entities accurately. We show that the model works well for both nested and flat NER through evaluation on 8 corpora and achieving SoTA performance on all of them, with accu- racy gains of up to 2.2 percentage points. 1 Introduction ‘Nested Entities’ are named entities containing ref- erences to other named entities as in [Bank of [China]], in which both [China] and [Bank of China] are named entities. Such nested entities are frequent in data sets like ACE 2004, ACE 2005 and GENIA (e.g., 17% of NEs in GENIA are nested (Finkel and Manning, 2009), altough the more widely used set such as CONLL 2002, 2003 and ONTONOTES only contain so called flat named en- tities and nested entities are ignored. The current SoTA models all adopt a neural net- work architecture without hand-crafted features, which makes them more adaptable to different tasks, languages and domains (Lample et al., 2016; Chiu and Nichols, 2016; Peters et al., 2018; De- vlin et al., 2019; Ju et al., 2018; Sohrab and Miwa, 2018; Strakov´a et al., 2019). In this paper, we in- troduce a method to handle both types of NEs in one system by adopting ideas from the biaffine de- pendency parsing model of Dozat and Manning (2017). For dependency parsing, the system pre- dicts a head for each token and assigns a relation to the head-child pairs. In this work, we reformu- late NER as the task of identifying start and end indices, as well as assigning a category to the span defined by these pairs. Our system uses a biaffine model on top of a multi-layer BiLSTM to assign scores to all possible spans in a sentence. After that, instead of building dependency trees, we rank the candidate spans by their scores and return the top-ranked spans that comply with constraints for flat or nested NER. We evaluated our system on three nested NER benchmarks (ACE 2004, ACE 2005, GENIA) and five flat NER corpora (CONLL 2002 (Dutch, Spanish) CONLL 2003 (English, Ger- man), …

Content truncated to 500 words