New low in reviewing quality:
R3:"why didn't you test your method on dataset X and Y?"
-> results for datasets X and Y are in the result table
I mean, it doesn't even require one to have read the paper, just a look at the tables and figures...
(this is at KDD btw)