Other dataset codes in the paper

[ljhry]

I am very interested in your work, if I want to train in FB15K and FB15K-237, what should I do Thank you very much

[chocolate9624]

Thanks for you interest!

Our experiments are based on this repository. So you can get the results by taking the following steps:

1, Add score function of PairRE to the repository;

2, Reproduce the results of FB15k-237 by this command: 'bash run.sh train PairRE FB15k-237 0 0 1024 256 1500 6.0 1.0 0.00005 100000 16 -dr'

3, Reproduce the results of FB15k by this command: 'bash run.sh train PairRE FB15k 0 0 1024 256 1500 15.0 1.0 0.0001 150000 16 -dr'

[laixiaolian]

Hi, I can not find DB100k and Sports datasets in your github, can you share these datasets? If I want to train in these datasets, what should I do? Looking forward to your reply！

[chocolate9624]

The DB100K dataset can be found here.
The command for reproduce the results is 'bash run.sh train PairRE DB100K 0 0 1024 256 500 9.0 1.0 0.00005 150000 16 -dr'

You can get the Sports dataset by sending mail to the author of 'Improved knowledge graph embedding using background taxonomic information'.
The command for reproduce the results is 'bash run.sh train PairRE Sports 0 0 128 16 200 3.0 1.0 0.0001 20000 16 -dr'

Apologize for the inconvenience of without uploading all the datasets.

[laixiaolian]

Thanks so much! Can you share the command to run the Sports dataset?

[chocolate9624]

The command is added here.

[laixiaolian]

I will try later, thank you so much!