GEMNET: Effective Gated Gazetteer Representations for Recognizing Complex Entities in Low-context Input

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1499–1512 June 6–11, 2021. ©2021 Association for Computational Linguistics 1499 GEMNET: Effective Gated Gazetteer Representations for Recognizing Complex Entities in Low-context Input Tao Meng∗ UCLA Los Angeles, USA Anjie Fang Amazon.com, Inc. Seattle, WA, USA Oleg Rokhlenko Amazon.com, Inc. Seattle, WA, USA Shervin Malmasi Amazon.com, Inc. Seattle, WA, USA tmeng@cs.ucla.edu {njfn,olegro,malmasi}@amazon.com Abstract Named Entity Recognition (NER) remains dif- ficult in real-world settings; current challenges include short texts (low context), emerging en- tities, and complex entities (e.g. movie names). Gazetteer features can help, but results have been mixed due to challenges with adding ex- tra features, and a lack of realistic evalua- tion data. It has been shown that including gazetteer features can cause models to overuse or underuse them, leading to poor generaliza- tion. We propose GEMNET, a novel approach for gazetteer knowledge integration, including (1) a flexible Contextual Gazetteer Represen- tation (CGR) encoder that can be fused with any word-level model; and (2) a Mixture-of- Experts gating network that overcomes the fea- ture overuse issue by learning to conditionally combine the context and gazetteer features, in- stead of assigning them fixed weights. To com- prehensively evaluate our approaches, we cre- ate 3 large NER datasets (24M tokens) reflect- ing current challenges. In an uncased setting, our methods show large gains (up to +49% F1) in recognizing difficult entities compared to existing baselines. On standard benchmarks, we achieve a new uncased SOTA on CoNLL03 and WNUT17. 1 Introduction Identifying entities is a core NLP task. Named Entity Recognition (NER) is the task of finding entities and recognizing their type (e.g., person or location). Mention Detection (MD) is a simpler task of identifying entity spans, without the types. Advances in neural NER have produced high scores on benchmark datasets like CoNLL03 and OntoNotes (Devlin et al., 2019). However, a num- ber of challenges remain. As noted by Augenstein et al. (2017), these scores are driven by the use of well-formed news text, the presence of “easy” enti- ties, and memorization effects due to entity overlap between train/test sets; these models perform sig- nificantly worse on unseen entities or noisy text. ∗This research was done during an internship at Amazon. 1.1 Current NER Challenges Beyond news text, many challenges remain in NER. Context information has been shown to be important for NER (Jayarao et al., 2018), and short texts like search queries are very challenging due to low context and a lack of surface features (Guo et al., 2009; Carmel et al., 2014). Unseen and emerging entities also pose a challenge (Bernier-Colborne and Langlais, 2020). Finally, some entities, like movie names are not simple noun phrases and are harder to recognize (Ashwini and Choi, 2014). Table 1 lists more details about these challenges, and how they can be evaluated. Entity Knowledge is essential for overcoming these issues, and critical in the absence of casing. Even a human may not correctly parse “what …

Content truncated to 500 words