Split data in training, validation and testing datasets

[sshivam95]

The overall data should be divided into three datasets:

• Training: contains 80% of the overall dataset
• Testing: contains 10% of overall dataset
• Validation: contains the remaining 10 5 of the dataset

[sshivam95]

This bash script does the trick. It splits the datasets using 3 CPU processors in parallel.

[sshivam95]

• Train data 45,782,638,889 triples
• Test data 5,722,829,163 triples
• Validation data 5,683,957,916 triples