News for hackers. Stuff that matters.
Re:Techniques for dilution, misdirection and control of a in... by chestervonwinch on Thursday November 12, 2015 @02:13PM (#10555410)
> I think a new killer app would be a form of automated shill-detection for forums... the trouble would be reducing the false positives to zero.

An interesting idea for AI! Although, the domain knowledge and contextual information required to classify a poster as say, "normal vs. astroturfer", might be too great, and vary from topic to topic. Also, there's the question of how positive / negative training examples would be obtained - who's going to admit to the ground-truth labels for the "astroturf" class, for example?

There are no comments on this comment yet.