The code of our paper "Improving Aspect Sentiment Triplet Extraction with Perturbed Masking and Edge-Enhanced Sentiment Graph Attention Network" accepted by IJCNN 2023.
-
python==3.7.6
-
torch==1.4.0
-
transformers==3.4.0
-
argparse==1.1
To train the model, run:
cd ./code
sh run.sh
or
python main.py --mode train --dataset res14 --batch_size 16 --epochs 100 --model_dir savemodel/ --seed 1000 --pooling avg --prefix ../data/D2/
The basic code framework is based on "https://github.com/CCChenhao997/EMCGCN-ASTE", thanks for the contribution of its open source code.