Source code of paper: "Online Opinion Conflict Interaction Recognition Based on Dependent Multi-Task Deep Learning". Accepted by ICWSM 2026 (International AAAI Conference on Web and Social Media).
Part of code (particularly data processing and loading) is based on the [Multi-architecture Open-source Chinese Text Classification] (https://github.com/649453932/Chinese-Text-Classification-Pytorch) project and has undergone significant modifications and adjustments.
ScienceNet_dataset and zhihu_dataset are two benchmark datasets we have constructed, which have been divided into train/dev/test sets.
We crawled users' opinion interaction records and manually annotated them.
-
Python 3.10
-
PyTorch 2.1.0
-
RTX 3090 GPU
-
CUDA 12.1
Recommended to use GPU for training (such as Google Colab, Kaggle, AutoDL, Aliyun, etc.)
The CIMDL.py file located in the models folder is the model file, which contains the definition scheme of the model.
The run.py, train_eval.py, and utils.py files are responsible for the project's operation, training, and processing respectively.
Additionally, the pre-trained model we used was downloaded from: https://huggingface.co/hfl/chinese-roberta-wwm-ext and https://huggingface.co/DMetaSoul/sbert-chinese-qmc-domain-v1.
python run.py --model CIMDL