Paper for 2024
This is the code of PipEA introduced in our paper, which is based on PEEA encoder.
The dataset can be unzipped by data.zip in your root.
The environment is provided in requirements.txt.
We also provide faster large-scale strategy for 100K and larger datasets. If you want to apply PipEA on large-scale KG, you can modify and run test.py
You can also apply 15K datasets based on the test.py, which only needs to generate embeddings firstly. The test.py will faster than original code.
Notably, the application on the large-scale KG need tf 2.x, which is different from original environment.
Our large scale environment is: