Awesome

Multi-modal Siamese Network for Entity Alignment

Codes for the paper "Multi-modal Siamese Network for Entity Alignment" published in Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD'2022).

Dataset

Three public multi-modal knowledge graphs with relational, attribute and visual knowledge from paper "MMKG: Multi-Modal Knowledge Graphs", i.e., FB15K, DB15K and YG15K. There are sameAs links between FB15K and DB15K as well as between FB15K and YG15K, which could be regarded as alignment relations. Please click here to download the original data and processed data. Permission for processed data needs to be requested on the link.

[Must Read!] Applicants should provide their name and organization. Once reviewed, we will approve it as soon as possible.

Code

Our codes were modified based on the public benchmark OpenEA. We appreciate the authors for making OpenEA open-sourced.

Citation

If you use this model or code, please kindly cite it as follows:

@inproceedings{chen2022multi,
  title={Multi-modal Siamese Network for Entity Alignment},
  author={Chen, Liyi and Li, Zhi and Xu, Tong and Wu, Han and Wang, Zhefeng and Yuan, Nicholas Jing and Chen, Enhong},
  booktitle={Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining},
  pages={118--126},
  year={2022}
}