I follow lots of pop culture feeds that recap TV shows that I don’t watch - so I train those recaps out of Newsblur. But fairly often I notice that items that hit on those trained phrases (and don’t appear to otherwise trigger positive or negative training) still appear in my feed.
As an example, in the Vulture feed (http://feeds.feedburner.com/nymag/vulture), I have trained with a thumbs down innumerable versions of “[Show Title] RECAP” in the title. But I noticed today that an item titled “Grey’s Anatomy Recap: Good Grief” was still appearing in my stream, despite correctly triggering my thumbs-down training on it (see screenshot):
In the sidebar the item doesn’t have a red dot on it, so somewhere there must be another factor that is countering the thumbs-down training, but I can’t figure out what it might be.
I see this most often for Vulture but it’s definitely happened for other feeds as well (and not just for pop culture recaps).
I’m not sure if this is carrying over to the iOS app, but I believe it is.