This repository contains data and code for the ACL23 (findings) paper: DiaASQ: A Benchmark of Conversational Aspect-based Sentiment Quadruple Analysis
Also see the project page for more details.
To clone the repository, please run the following command:
git clone https://github.com/unikcc/DiaASQ
📢 2023-05-10
: Released code and dataset.
⚡ 2022-12-10
: Created repository.
In this work, we propose a new task named DiaASQ, which aims to extract Target-Aspect-Opinion-Sentiment quadruples from the given dialogue. More details about the task can be found in our paper.
The model is implemented using PyTorch. The versions of the main packages:
- python>=3.7
- torch>=1.8.1
Install the other required packages:
pip install -r requirements.txt
-
Dataset: the dataset can be found at:
data/dataset
-
Train && Evaluate on the Chinese dataset
bash scripts/train_zh.sh
-
Train && Evaluate on the English dataset
bash scripts/train_en.sh
-
GPU memory requirements
Dataset Batch size GPU Memory Chinese 2 8GB. English 2 16GB. -
Customized hyperparameters:
You can set hyperparameters inmain.py
orsrc/config.yaml
, and the former has a higher priority.
If you use our dataset, please cite the following paper:
@article{lietal2022arxiv,
title={DiaASQ: A Benchmark of Conversational Aspect-based Sentiment Quadruple Analysis},
author={Bobo Li, Hao Fei, Fei Li, Yuhan Wu, Jinsong Zhang, Shengqiong Wu, Jingye Li, Yijiang Liu, Lizi Liao, Tat-Seng Chua, Donghong Ji}
journal={arXiv preprint arXiv:2211.05705},
year={2022}
}