CrossLink: Enhancing Cross-domain Link Prediction via Evolution Process Modeling

WWW 2025

🚀 Introduction

CrossLink learns the evolution pattern of a specific downstream graph and subsequently makes pattern-specific link predictions. It employs a technique called conditioned link generation, which integrates both evolution and structure modeling to perform evolution-specific link prediction. This conditioned link generation is carried out by a transformer-decoder architecture, enabling efficient parallel training and inference. CrossLink is trained on extensive dynamic graphs across diverse domains, encompassing 6 million dynamic edges. Extensive experiments on eight untrained graphs demonstrate that CrossLink achieves state-of-the-art performance in cross-domain link prediction. Compared to advanced baselines under the same settings, CrossLink shows an average improvement of 11.40% in Average Precision across eight graphs. Impressively, it surpasses the fully supervised performance of 8 advanced baselines on 6 untrained graphs.

🛠️ Prerequisites

Environment

conda create -n your_env_name python=3.8
conda activate crosslink
pip install -r requirements.txt

Dataset

Please keep the dataset in the fellow format:

Unnamed: 0	u	i	ts	label	idx
idx-1	source node	target node	interaction time	defalut: 0	from 1 to the #edges

You can prepare those data by the code in preprocess_data folder

You can also use our processed data in huggingface

💡 Usage

Training

Evaluation

📅 TODO

• Release evaluation code and checkpoint
• Release training code

💞 Acknowledgment

Our code is built refer to DyGLib

📚 Citation

If you find this work helpful, please consider citing:

@misc{huang2024graphmodelcrossdomaindynamic,
  title={One Graph Model for Cross-domain Dynamic Link Prediction}, 
  author={Xuanwen Huang and Wei Chow and Yang Wang and Ziwei Chai and Chunping Wang and Lei Chen and Yang Yang},
  year={2024},
  eprint={2402.02168},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  url={https://arxiv.org/abs/2402.02168}, 
}