brown-q / dsea

DSEA: Leveraging multimodal features for knowledge graph entity alignment based on dynamic self-attention networks

Datasets

Please download the datasets here and extract them into root directory.
ent_ids_1: entity ids in the source KG
ent_ids_2: entity ids in the target KG
ref_ent_ids: entity alignment, list of pairs like (e_s \t e_t)
triples_1: relation triples in the source KG
triples_2: relation triples in the target KG

Environment

python==3.6.13
networkx==2.5.1
apex==0.1
pytorch==1.7.1
torch_geometric==2.0.2
cuda==11.2

Running

CUDA_VISIBLE_DEVICES=0 python train.py --data data/DBP15K --lang ja_en