Would it be possible to create a bayesian spam filter-like system where new users edits are analyzed and matched to old accounts? That way, admins could get a h

Would it be possible to create a bayesian spam filter-like system where new users edits are analyzed and matched to old accounts? That way, admins could get a heads up for a new user that is likely to be a pseudonym of an old user. Would it be technically feasable? Maybe there are social consequences to this as well? What about false postives?