GTS

Grid Tagging Scheme for Aspect-oriented Fine-grained Opinion Extraction. Zhen Wu, Chengcan Ying, Fei Zhao, Zhifang Fan, Xinyu Dai, Rui Xia. In Findings of EMNLP, 2020.

Evaluation on ASOTE

We folk this repository from the GTS repository and evaluate the GTS model on the ASOTE task.

Supported datasets

res14, lap14, res15, res16

LSTM

sh run.sh 4 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model bilstm --dataset res14 > bilstm.res14.log 2>&1 & sh run.sh 4 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model bilstm --dataset res14 > bilstm.res14.log 2>&1 &

sh run.sh 5 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model bilstm --dataset lap14 > bilstm.lap14.log 2>&1 & sh run.sh 5 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model bilstm --dataset lap14 > bilstm.lap14.log 2>&1 &

sh run.sh 4 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model bilstm --dataset res15 > bilstm.res15.log 2>&1 & sh run.sh 4 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model bilstm --dataset res15 > bilstm.res15.log 2>&1 &

sh run.sh 5 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model bilstm --dataset res16 > bilstm.res16.log 2>&1 & sh run.sh 5 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model bilstm --dataset res16 > bilstm.res16.log 2>&1 &

CNN

sh run.sh 6 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model cnn --dataset res14 > cnn.res14.log 2>&1 & sh run.sh 6 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model cnn --dataset res14 > cnn.res14.log 2>&1 &

sh run.sh 7 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model cnn --dataset lap14 > cnn.lap14.log 2>&1 & sh run.sh 7 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model cnn --dataset lap14 > cnn.lap14.log 2>&1 &

sh run.sh 6 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model cnn --dataset res15 > cnn.res15.log 2>&1 & sh run.sh 6 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model cnn --dataset res15 > cnn.res15.log 2>&1 &

sh run.sh 7 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode train --model cnn --dataset res16 > cnn.res16.log 2>&1 & sh run.sh 7 src/NNModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/ --task triplet --mode test --model cnn --dataset res16 > cnn.res16.log 2>&1 &

BERT

sh run.sh 0 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode train --dataset res14 > bert.res14.log 2>&1 & sh run.sh 0 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode test --dataset res14 > bert.res14.log 2>&1 &

sh run.sh 1 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode train --dataset lap14 > bert.lap14.log 2>&1 & sh run.sh 1 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode test --dataset lap14 > bert.lap14.log 2>&1 &

sh run.sh 2 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode train --dataset res15 > bert.res15.log 2>&1 & sh run.sh 2 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode test --dataset res15 > bert.res15.log 2>&1 &

sh run.sh 3 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode train --dataset res16 > bert.res16.log 2>&1 & sh run.sh 3 src/BertModel/main.py --prefix data/ --model_dir GTS-ASOTE-model/bert/ --task triplet --mode test --dataset res16 > bert.res16.log 2>&1 &

Data

[Data] [Pre-trained Model(from huggingface)]. Data format descriptions are here.

Note: our opinion triplet datasets are from alignments of our previous work TOWE datasets and the original SemEval 2014, 2015, 2016 datasets.

Requirements

See requirement.txt or Pipfile for details

pytorch==1.7.1
transformers==3.4.0
python=3.6

Usage

Training

For example, you can use the folowing command to fine-tune Bert on the OPE task (the pre-trained Bert model is saved in the folder "pretrained/"):

python main.py --task pair --mode train --dataset res14

The best model will be saved in the folder "savemodel/".

Testing

For example, you can use the folowing command to test Bert on the OPE task:

python main.py --task pair --mode test --dataset res14

Note: In our pre-experiments, a smaller batch size and learning rate can achieve better performance on certain datasets, while we use a general setting in our paper to save time instead of adopting grid search.

Citation

If you used the datasets or code, please cite our paper:

@inproceedings{wu-etal-2020-grid,
    title = "Grid Tagging Scheme for Aspect-oriented Fine-grained Opinion Extraction",
    author = "Wu, Zhen  and
      Ying, Chengcan  and
      Zhao, Fei  and
      Fan, Zhifang  and
      Dai, Xinyu  and
      Xia, Rui",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2020",
    month = nov,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.findings-emnlp.234",
    doi = "10.18653/v1/2020.findings-emnlp.234",
    pages = "2576--2585",
}

Reference

[1]. Zhen Wu, Chengcan Ying, Fei Zhao, Zhifang Fan, Xinyu Dai, Rui Xia. Grid Tagging Scheme for Aspect-oriented Fine-grained Opinion Extraction. In Findings of EMNLP, 2020.

[2]. Zhifang Fan, Zhen Wu, Xin-Yu Dai, Shujian Huang, Jiajun Chen. Target-oriented Opinion Words Extraction with Target-fused Neural Sequence Labeling. In Proceedings of NAACL, 2019.

hoangdzung/GTS-ASOTE

GTS