Augmenting Low-Resource Text Classification with Graph-Grounded Pre-training and Prompting

We provide the implementation of G2P2 model, which is the source code for the SIGIR 2023 paper "Augmenting Low-Resource Text Classification with Graph-Grounded Pre-training and Prompting".

The repository is organised as follows:

• dataset/: the directory of data sets. Currently, it only has the dataset of Cora, if you want the three processed Amazon datasets, you can download and put them under this directory, the link is https://drive.google.com/drive/folders/1IzuYNIYDxr63GteBKeva-8KnAIhvqjMZ?usp=sharing. Besides, this link also contains the 4 pre-trained models, under the directory of "pre-trained model". The web site of the original Cora dataset: https://people.cs.umass.edu/~mccallum/data/cora-classify.tar.gz , the web site of the three raw Amazon datasets: http://deepyeti.ucsd.edu/jianmo/amazon/index.html.
• res/: the directory of saved models.
• bpe_simple_vocab_16e6.txt.gz: vocabulary for simple tokenization.
• data.py, data_graph.py: for data loading utilization.
• main_test.py, main_test_amazon.py: testing entrance for cora, testing entrance for Amazon datasets.
• main_train.py, main_train_amazon.py: pre-training entrance for cora, pre-training entrance for Amazon datasets.
• model.py, model_g_coop.py: model for pre-training, model for prompt tuning.
• multitask.py, multitask_amazon.py: task generator for cora, task generator for Amazon datasets.
• requirements.txt: the required packages.
• simple_tokenizer: a simple tokenizer.

For pre-train:

On Cora dataset,

python main_train.py 

If on Amazon datasets, it should be:

python main_train_amazon.py

For prompt tuning and testing:

On Cora dataset,

python main_test.py 

If on Amazon datasets, it should be:

python main_test_amazon.py

For zero-shot testing:

First, change directory to /zero-shot

On Cora dataset,

python zero-shot-cora.py 

If on Amazon datasets, it should be:

python zero-shot-amazon.py

Cite

@inproceedings{DBLP:conf/sigir/Wen023,
  author       = {Zhihao Wen and
                  Yuan Fang},
  title        = {Augmenting Low-Resource Text Classification with Graph-Grounded Pre-training
                  and Prompting},
  booktitle    = {Proceedings of the 46th International {ACM} {SIGIR} Conference on
                  Research and Development in Information Retrieval, {SIGIR} 2023, Taipei,
                  Taiwan, July 23-27, 2023},
  pages        = {506--516}
}