WWW24-ID259-Span-Pair Interaction and Tagging for Dialogue-Level Aspect-Based Sentiment Quadruple Analysis

Environment

• conda create -n mrm python==3.8

• conda activate mrm

• [CUDA 11.1] pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https://download.pytorch.org/whl/torch_stable.html

• pip install -r requirements.txt

Data

We conduct experiments on Chinese and English datasets. The datasets are located in data/diaasq.

Running

1. training a Span Ranker and saving the checkpoints to data/saves/ranker.

• cd span_ranker/src
• bash run_train.sh

2. generating span indices and selecting the top K span by the trained span ranker, saving the .pkl files to data/saves/ranker.

• cd span_ranker/src
• bash run_eval.sh

3. training a MRM and obtain the results.

• cd mrm/src
• bash run.sh