Named Entity Recognition in the Style of Object Detection

TL;DR

This paper introduces a novel two-stage object detection-inspired approach for nested named entity recognition, leveraging region proposals and specialized loss functions to improve entity detection accuracy.

Contribution

It adapts the object detection framework to NER, especially nested entities, with a new loss function and demonstrates competitive results on multiple datasets.

Findings

Comparable performance on flat NER tasks with traditional models

Achieved 85.6% F1 on ACE2005 for nested NER

Adding random regions improves precision

Abstract

In this work, we propose a two-stage method for named entity recognition (NER), especially for nested NER. We borrowed the idea from the two-stage Object Detection in computer vision and the way how they construct the loss function. First, a region proposal network generates region candidates and then a second-stage model discriminates and classifies the entity and makes the final prediction. We also designed a special loss function for the second-stage training that predicts the entityness and entity type at the same time. The model is built on top of pretrained BERT encoders, and we tried both BERT base and BERT large models. For experiments, we first applied it to flat NER tasks such as CoNLL2003 and OntoNotes 5.0 and got comparable results with traditional NER models using sequence labeling methodology. We then tested the model on the nested named entity recognition task ACE2005 and Genia, and got F1 score of 85.6$\%$ and 76.8$\%$ respectively. In terms of the second-stage training, we found that adding extra randomly selected regions plays an important role in improving the precision. We also did error profiling to better evaluate the performance of the model in different circumstances for potential improvements in the future.