It's not really supervised learning. I suppose one could argue that the bootstrapping phase resembles supervised learning, since you are in fact measuring how w

It's not really supervised learning. I suppose one could argue that the bootstrapping phase resembles supervised learning, since you are in fact measuring how well each user discriminates between honeypots and acceptable articles. However, after that it's more like unsupervised learning. If I were going to use this terminology, I would say that implicit honeypots are a generative model that is bootstrapped via a discriminative learning phase. And who is being fancy? It's not like those formulas are that confusing, are they? :)