GAGAT model dataset

The GAGAT model is implemented using the Pytorch framework on an NVIDIA GeForce GTX 1080 Ti GPU with 11GB of memory (Pytorch 2.3.0; Cuda 11.8; Python 3.9). The specific model dataset consists of three parts: the original dataset (FB15K-237, WN18RR, and four sub datasets FB15k-237v1, FB15k-237v2, FB15k-237v3, FB15k-237v4 of FB15K-237), and the TransE pre-processing model; And the GAGAT model ontology. After configuring the relevant environment, modify the address paths of all models to the specified original dataset. First, initialize the preprocessing of the original dataset using the TransE preprocessing model, generate implicit information and entity machine relationship embeddings using the implicit global information vector generation program, and finally input the embeddings into the encoder of the GAGAT model for training and evaluation.