RaNER

Retrieval-Augmented Named Entity Recognition

RaNER is a re-implementation of our ACL-IJCNLP 2021 paper: Improving Named Entity Recognition by External Context Retrieving and Cooperative Learning [github]

RaNER is a framework for improving the accuracy of NER models through retrieving external contexts, then use the cooperative learning approach to improve the both input views. It can be illustrated as follows:

Prepare dataset

After we retrieve the external contexts, we can simply concat them to the original sentences. The label used for the contexts should be X.

for example:

EU B-ORG
rejects O
German B-MISC
call O
to O
boycott O
British B-MISC
lamb O
. O
<EOS> X
EU X
officials X
sought X
in X
vain X

Training a new model

python -m scripts.train -c examples/raner/configs/wnut17.yaml

Benchmarks

MultiCoNER

Dataset	Baseline-F1	RaNER-F1	Modelcard & Demo
MultiCoNER-BN	82.69	85.11	ModelScope
MultiCoNER-DE	91.71	95.0	ModelScope
MultiCoNER-EN	88.70	96.59	ModelScope
MultiCoNER-ES	86.54	94.64	ModelScope
MultiCoNER-FA	81.85	95.97	ModelScope
MultiCoNER-HI	83.13	85.28	ModelScope
MultiCoNER-KO	86.25	95.49	ModelScope
MultiCoNER-NL	89.92	97.28	ModelScope
MultiCoNER-RU	81.52	95.14	ModelScope
MultiCoNER-TR	88.52	97.83	ModelScope
MultiCoNER-ZH	85.43	91.44	ModelScope
Baseline indicates Transformer-CRF model with the same pretrained backbone.

Citation

@inproceedings{wang2021improving,
    title = "{{Improving Named Entity Recognition by External Context Retrieving and Cooperative Learning}}",
    author={Wang, Xinyu and Jiang, Yong and Bach, Nguyen and Wang, Tao and Huang, Zhongqiang and Huang, Fei and Tu, Kewei},
    booktitle = "{the Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (\textbf{ACL-IJCNLP 2021})}",
    month = aug,
    year = "2021",
    publisher = "Association for Computational Linguistics",
}