Hmm, curious. I suppose I would wonder, then, if there's no dataset how any algorithm could be developed. What I mean is ... once you've developed something, how do you test how well it performs? How do you analyze the effectiveness of the algorithm, the false positive rate, etc.