Different number of relations in the datasets vs reported number in the paper

[weihangzhang]

Hi,

When reading your the knowledge graph data in each of the language, I am getting different numebr of relations for each of the language v.s. what has been reported in your paper.

num relations I got:
EL 160, JA 172, ES 241, FR 236, EN 861
reported num relations in paper:
EL 111, JA 128, ES 144, FR 178, EN 831

I include all train, validation and test triples when counting relations. Numbers are still different if I exclude the test triples. The total number of triples match the reported number in the paper, but number of relation is different.
Could you please suggest if I am missing something here?