hccngu / meta-sn

Meta-Learning Siamese Network for Few-Shot TextClassification

This repository contains the code and data for our DASFAA 2023 paper:

Meta-Learning Siamese Network for Few-Shot TextClassification

If you find this work useful and use it on your own research, please cite our paper.

@inproceedings{han2023meta,
  title={Meta-learning Siamese Network for Few-Shot Text Classification},
  author={Han, Chengcheng and Wang, Yuhe and Fu, Yingnan and Li, Xiang and Qiu, Minghui and Gao, Ming and Zhou, Aoying},
  booktitle={Database Systems for Advanced Applications: 28th International Conference, DASFAA 2023, Tianjin, China, April 17--20, 2023, Proceedings, Part III},
  pages={737--752},
  year={2023},
  organization={Springer}
}

Data

We ran experiments on a total of 6 datasets. You may unzip our processed data file data.zip and put the data files under data/ folder.

Dataset	Notes
20 Newsgroups (link)	Processed data available. We used the 20news-18828 version, available at the link provided.
Reuters-21578 (link)	Processed data available.
Amazon reviews (link)	We used a subset of the product review data. Processed data available.
HuffPost headlines (link)	Processed data available.
RCV1 (link)	Processed data available.
FewRel (link)	Processed data available.

Please download pretrained word embedding file wiki.en.vec from here and put it under pretrain_wordvec/ folder.

Quickstart

After you have finished configuring the data/ folder and the pretrain_wordvec/ folder, you can run our model with the following commands.

cd bin
sh 1-shot.sh

or

cd bin
sh 5-shot.sh

You can also adjust the model by modifying the parameters in the 1-shot.sh or 5-shot.sh file to run on diffrent data sets.

Dependencies

• Python 3.7
• PyTorch 1.6.0
• numpy 1.18.5
• torchtext 0.7.0
• termcolor 1.1.0
• tqdm 4.46.0
• CUDA 10.2

meta-sn's Issues

I'm curious if this is a bug

Hello, during my study of your code, I noticed that in the return values of loader.load_dataset(), was "val_data" mistakenly written as "test_data"?

how to use BERT as encoder?

I try to run your model for my research.
for fair comparison, I try to use BERT as encoder. In your paper, you also did an experiment of BERT as encoder.
BUT, I cannot find how to use BERT as encoder for your model.

I found that an argument (--bert) in shell script. however, I cannot find any source code to use BERT as an encoder in the source.