$k$NN-NER: Named Entity Recognition with Nearest Neighbor Search
This work addresses long-tail and few-shot learning challenges in NER for NLP practitioners, representing an incremental improvement over existing methods.
The paper tackles the problem of named entity recognition by introducing a k-nearest neighbor framework that augments entity label distributions, achieving a new state-of-the-art F1-score of 72.03 (+1.25 improvement) on the Chinese Weibo dataset and showing comparable results with 40% less training data.
Inspired by recent advances in retrieval augmented methods in NLP~\citep{khandelwal2019generalization,khandelwal2020nearest,meng2021gnn}, in this paper, we introduce a $k$ nearest neighbor NER ($k$NN-NER) framework, which augments the distribution of entity labels by assigning $k$ nearest neighbors retrieved from the training set. This strategy makes the model more capable of handling long-tail cases, along with better few-shot learning abilities. $k$NN-NER requires no additional operation during the training phase, and by interpolating $k$ nearest neighbors search into the vanilla NER model, $k$NN-NER consistently outperforms its vanilla counterparts: we achieve a new state-of-the-art F1-score of 72.03 (+1.25) on the Chinese Weibo dataset and improved results on a variety of widely used NER benchmarks. Additionally, we show that $k$NN-NER can achieve comparable results to the vanilla NER model with 40\% less amount of training data. Code available at \url{https://github.com/ShannonAI/KNN-NER}.