Boundary Smoothing for Named Entity Recognition

ACL 2022  ·  Enwei Zhu, Jinpeng Li ·

Neural named entity recognition (NER) models may easily encounter the over-confidence issue, which degrades the performance and calibration. Inspired by label smoothing and driven by the ambiguity of boundary annotation in NER engineering, we propose boundary smoothing as a regularization technique for span-based neural NER models. It re-assigns entity probabilities from annotated spans to the surrounding ones. Built on a simple but strong baseline, our model achieves results better than or competitive with previous state-of-the-art systems on eight well-known NER benchmarks. Further empirical analysis suggests that boundary smoothing effectively mitigates over-confidence, improves model calibration, and brings flatter neural minima and more smoothed loss landscapes.

PDF Abstract ACL 2022 PDF ACL 2022 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Nested Named Entity Recognition ACE 2004 Baseline + BS F1 87.98 # 8
Nested Named Entity Recognition ACE 2005 Baseline + BS F1 87.15 # 5
Named Entity Recognition (NER) CoNLL 2003 (English) Baseline + BS F1 93.65 # 12
Chinese Named Entity Recognition MSRA Baseline + BS F1 96.26 # 3
Chinese Named Entity Recognition OntoNotes 4 Baseline + BS F1 82.83 # 3
Named Entity Recognition (NER) Ontonotes v5 (English) Baseline + BS F1 91.74 # 3
Chinese Named Entity Recognition Resume NER Baseline + BS F1 96.66 # 3
Chinese Named Entity Recognition Weibo NER Baseline + BS F1 72.66 # 2

Methods