fanshu6hao/heproto

The source code for paper: HEProto: A Hierarchical Enhancing ProtoNet based on Multi-Task Learning for Few-shot Named Entity Recognition

HEProto

 

The source code for paper:

HEProto: A Hierarchical Enhancing ProtoNet based on Multi-Task Learning for Few-shot Named Entity Recognition (full paper of CIKM '23)

If you find our work useful for your research, please cite the following paper:

@inproceedings{chen2023heproto,
  author       = {Wei Chen and
                  Lili Zhao and
                  Pengfei Luo and
                  Tong Xu and
                  Yi Zheng and
                  Enhong Chen},
  title        = {HEProto: {A} Hierarchical Enhancing ProtoNet based on Multi-Task Learning
                  for Few-shot Named Entity Recognition},
  booktitle    = {{CIKM}},
  pages        = {296--305},
  publisher    = {{ACM}},
  year         = {2023}
}

Overview

File Structure

-HEProto
    -data
    -dataset
    -debug
    -outputs
    -scripts
    ...

Dataset

We use the Few-NERD Arxiv V6 Version dataset, which you can download from the website (Sampled Datasets) and put in the ./dataset directory.

Quick Start

Train

bash script/run.sh

Eval

bash script/pred.sh

Results

Few-NERD (Inter)

Few-NERD (Intra)

Acknowledge

Our code is based on DecomposedMetaNER, thanks very much for their great work!