question about formula (2) and (5)

[fanyangxyz]

question about formula (2) and (5)

[fanyangxyz]

We are not intended to say that Eq (2) and Eq (5) are equivalent, but that rules parameterized using Eq (5) (or (6)-(8)) can be converted back to Eq (2), which is essentially what Algorithm 1 is about. Hope this helps.

[chenhch8]

Thank you for your reply!
I have another question. I tried to recover the experimental results on KG completion on the WN18 dataset, but the best hit @10 is 84.7, which is much smaller than the 94.5 shown in your paper. I also changed some parameters, but it just improved a bit. Do you mind telling me its parameter settings?

[fanyangxyz]

Did you use the evaluation steps in the README? My guess is that your number is not the filtered hit@10.