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

Abstract

Named Entity Recognition (NER) remains difficult in real-world settings; current challenges include short texts (low context), emerging entities, and complex entities (e.g. movie names). Gazetteer features can help, but results have been mixed due to challenges with adding extra features, and a lack of realistic evaluation data. It has been shown that including gazetteer features can cause models to overuse or underuse them, leading to poor generalization. We propose GEMNET, a novel approach for gazetteer knowledge integration, including (1) a flexible Contextual Gazetteer Representation (CGR) encoder that can be fused with any word-level model; and (2) a Mixture-of-Experts gating network that overcomes the feature overuse issue by learning to conditionally combine the context and gazetteer features, instead of assigning them fixed weights. To comprehensively evaluate our approaches, we create 3 large NER datasets (24M tokens) reflecting 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.

USTC-NELSLIP at SemEval-2022 Task 11: Gazetteer-Adapted Integration Network for Multilingual Complex Named Entity Recognition

“…In order to improve the performance of our models on low-context instances, a set of annotated sentences are generated from the MS-MARCO QnA corpus (V2.1) (Nguyen et al, 2016) and the ORCAS dataset (Craswell et al, 2020), which are mentioned in Meng et al (2021). Our trained models and existing NER systems (e.g., spaCy) are applied to identify entities in these corpora, and only templates identically recognized by all models have been reserved.…”

“…One of the classic approaches to solving this problem is to integrate external entity knowledge or gazetteers into neural architectures (Liu et al, 2019;Rijhwani et al, 2020;Meng et al, 2021). Typically, the two representations from a language model like BERT (Devlin et al, 2019) and a gazetteer network like BiLSTM (Hochreiter and Schmidhuber, 1997) respectively are combined as one merged embedding, which is further fed into a NER classifier such as a conditional random field (CRF) (Lafferty et al, 2001).…”

CMNEROne at SemEval-2022 Task 11: Code-Mixed Named Entity Recognition by leveraging multilingual data

“…The algorithms used were Decision tree, Long Short-Term Memory (LSTM), and Conditional Random Field (CRF). (Meng et al, 2021;Fetahu et al, 2021) presented a novel CM NER model. They proposed a gated architecture that enhances existing multilingual Transformers by dynamically infusing multilingual knowledge bases, a.k.a gazetteers.…”

LMN at SemEval-2022 Task 11: A Transformer-based System for English Named Entity Recognition

“…Complex named entities (e.g., titles of creative works) are typically not simple nouns and are harder to recognize. The challenges of NER for recognizing complex entities and in lowcontext situations was recently outlined by Meng et al (2021b). Other work has extended this to multilingual and code-mixed settings (Fetahu et al, 2021).…”

“…Recognizing complex named entities (NEs) is a challenging research problem, but it has not received sufficient attention from the natural language processing community (Meng et al, 2021a;Fetahu et al, 2021). Complex NEs can be complex noun phrases (e.g., National Baseball Hall of Fame and Museum), gerunds (e.g., Saving Private Ryan), infinitives (e.g., To Build a Fire), or even full clauses (e.g., I Capture The Castle).…”