Knowledge Graph Population via a Generative Fashion. The code is implemented by Python 3 and Keras
"CVAE_GAN_BERT_TransE.py" is the proposed algorithm.
"transformer.py" is a implementation of transformer layer.
"data" is the filefolder which contains experiment data.
The output of our algorithm is put in "computing_results/new".
"computing_results/ExplainResults_TransX.py" can translate the results from id to name. So you can read the results.
"computing_results/NotateGeneratedResults.py" can notate the generated triples as N(new), E(Existed), C(Need to Check), and give statistics
"computing_results/ComputeRankLoss.py" can compute the Label Ranking Average Precision of computing result