/tadner_test

Primary LanguagePythonMIT LicenseMIT

Type-Aware Decomposed Framework for Few-Shot Named Entity Recognition

PWC PWC

Code and data of our paper "Type-Aware Decomposed Framework for Few-Shot Named Entity Recognition" accepted by Findings of EMNLP 2023.

Paper link: Type-Aware Decomposed Framework for Few-Shot Named Entity Recognition

Overview

Framework of TadNER

1 Quick Start

Here we give an easy example for training and test on Domain-Transfer / FEW-NERD intra settings.

1.1 Environment

Python=3.8

pip install -r requirements.txt

1.2 train and test Domain Transfer CoNLL2003

bash run.sh

OR

bash run_fewnerd.sh

Note: Due to copyright restrictions, we apologize for not being able to provide some datasets in this repository. You can download FEW-NERD dataset at https://ningding97.github.io/fewnerd/, OntoNotes 5.0 at https://catalog.ldc.upenn.edu/LDC2013T19, I2B2 at https://portal.dbmi.hms.harvard.edu/projects/n2c2-nlp/. Here, for your convenience, we have released a small portion of the data as an example.

Acknowledge

The sampled few-shot data under Domain-Transfer settings is from https://github.com/psunlpgroup/CONTaiNER, thanks for their excellent work!