The Utility and Interplay of Gazetteers and Entity Segmentation for Named Entity Recognition in English

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 3990–4002 August 1–6, 2021. ©2021 Association for Computational Linguistics 3990 The Utility and Interplay of Gazetteers and Entity Segmentation for Named Entity Recognition in English Oshin Agarwal1 and Ani Nenkova1,2 1University of Pennsylvania 2Adobe Research oagarwal@seas.upenn.edu, nenkova@adobe.com Abstract Recent papers have introduced methods to in- corporate gazetteer features and entity segmen- tation techniques in neural named entity recog- nition models. These papers rely on differ- ent resources and include features not related to the use of gazetteers, rendering impossible the comparison of the relative effectiveness of the approaches. Here, we provide a compre- hensive overview of methods for incorporat- ing gazetteers and for entity segmentation. We evaluate representative methods from each in similar settings for a fair comparison and iden- tify the ones that are consistently better across datasets and input representations. We further show that gazetteers improve entity segmenta- tion and not just entity typing. Hence, we ex- plore their utility in recognizing long entities, a problem for which entity segmentation tech- niques were developed. Our work explains the mechanisms via which gazetteers improve the performance of neural NER models. 1 Introduction Named Entity Recognition (NER) has the unique property of being a task appealing to researchers and at the same time being fairly robust for imme- diate practical applications. In many domains, it is of interest to identify segments of text convey- ing a concept of a given type—a person (Grish- man and Sundheim, 1996), an event (Hovy et al., 2006), a disease (Do˘gan et al., 2014), a gene (Kim et al., 2003), a chemical (Krallinger et al., 2015), a food (Magnolini et al., 2019), an item of clothing (Putthividhya and Hu, 2011), a research technique (Augenstein et al., 2017a), etc. Approaches to NER are typically not domain- specific, treating the problem as a sequence la- belling task regardless of the categories of interest. Yet, researchers also widely agree that named entity recognition is a knowledge intensive task (Ratinov and Roth, 2009; Seyler et al., 2018): the availability of external knowledge resources in the form of lists of example entities of a given type, or gazetteers, improve performance almost universally. Since gazetteers are readily available, from knowledge bases, databases of products and specialized ontolo- gies, having practical guidance on how to handle gazetteers in NER would be valuable. In this paper, we provide a survey of how gazetteers have been used in neural approaches to NER in English and compare key approaches with the popular biLSTM-CRF architecture. To ensure that our conclusions accurately character- ize the utility of gazetteers, we test the approaches on several datasets from different genres covering newswire, conversations and twitter. The extensive head-to-head comparison reveals that while certain approaches are consistently beneficial, others are variable, impressively improving results on one dataset but reducing performance on others. Gazetteers typically contain multi-word entries. In contrast, the majority of entity mentions in text are single word, with lower performance of models on longer entities. This discrepancy highlights …

Content truncated to 500 words