Comment by jdup7

Yeah probably could have done better at describing the methodology. The dataset is just the confirmed (manually by a human) phishing urls. We only included the FPs to show that the tooling isn't perfect there were many TNs that we did not include. Going forward we could definitely frame these results better.