I have a dataset of ratings given to movies by users. I've applied the method mentioned in this paper to detect fake votes(I've used $H(X)$ and $M(X)$ measures). But the dataset I'm using doesn't have any label for checking the false alarm rate and detection rate of my code. So is there any way to report these two criteria without having labels?
Asked
Active
Viewed 16 times
1
-
Try semi-supervised models like label propagation to create labels. – fuwiak Apr 05 '20 at 16:52