This repository is the PyTorch implementation of our ACLT model in EMNLP 2021 Paper "To be Closer: Learning to Link up Aspects with Opinions".

The code has been tested in the following enviroment:

• Ubuntu 18.04.4 LTS
• Python 3.8.3
• torch 1.9.0
• cuda 10.2
• cudnn 7605

To install:

• conda create -n aclt python=3.8.3
• source activate aclt
• git clone https://github.com/zyxnlp/ACLT.git
• cd ACLT
• pip install -r requirements.txt

Pre-trained model:

• Downlaod bert-base model from here
• Unzip the .zip file to the folder ACLT/pretrain_model/

• Change the access mode of the .sh file by chmod 700 *.sh
• ./run.sh

If you find our work or the code useful, please consider cite our paper using:

@inproceedings{zhou2021closer,
 title={To be Closer: Learning to Link up Aspects with Opinions},
 author={Zhou, Yuxiang and Liao, Lejian and Gao, Yang and Jie, Zhanming and Lu, Wei},
 booktitle={Proceedings of EMNLP},
 year={2021}
}