Rethinking Generalization of Neural Models: A Named Entity Recognition Case Study
This work addresses the understanding of generalization limitations in neural NER models for NLP researchers, but it is incremental as it focuses on analysis and diagnosis rather than proposing new methods.
The paper analyzes the generalization behavior of neural models in Named Entity Recognition (NER), diagnosing bottlenecks through breakdown performance analysis, annotation errors, dataset bias, and category relationships, and provides datasets and resources for future research.
While neural network-based models have achieved impressive performance on a large body of NLP tasks, the generalization behavior of different models remains poorly understood: Does this excellent performance imply a perfect generalization model, or are there still some limitations? In this paper, we take the NER task as a testbed to analyze the generalization behavior of existing models from different perspectives and characterize the differences of their generalization abilities through the lens of our proposed measures, which guides us to better design models and training methods. Experiments with in-depth analyses diagnose the bottleneck of existing neural NER models in terms of breakdown performance analysis, annotation errors, dataset bias, and category relationships, which suggest directions for improvement. We have released the datasets: (ReCoNLL, PLONER) for the future research at our project page: http://pfliu.com/InterpretNER/. As a by-product of this paper, we have open-sourced a project that involves a comprehensive summary of recent NER papers and classifies them into different research topics: https://github.com/pfliu-nlp/Named-Entity-Recognition-NER-Papers.