afaik they do use text-based features in detection, but it's easy to come up with one off examples that will fail the spam filter,
-
-
-
As well as filters they surely have humans. And if I can surface a network without their internal tools, they should have already.
-
it's not as simple. For eg, would you suspend this person based on the text match?pic.twitter.com/3WBPMMPuOG
-
Of course not. You never do anything based on one feature. My point is they don't even need to try to hide.
-
yes, there is a ton of signal still waiting to be harvested. With deep learning,4eg all NSFW image features should be easy to detect
End of conversation
New conversation -
-
-
how do you know these are hacked accounts and not bots?
-
Some of the accounts go back to 2010.
-
oh. I saw some that have single digit number of tweets
-
They don't tweet much. They spam likes.
-
ah right. Now I know where all those likes for three year old tweets come from
End of conversation
New conversation -
-
-
could you possibly like, remove the bots and trolls from the net entirely?pic.twitter.com/wH1fHdW3vr
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I'm sure a lot of real accounts follow spam accounts for the pics - thus Twitter has a conflict of interest here.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Similar problems pop up on Facebook too. I'm starting to think they care more about post counts than really blocking spam.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Identical text can help detect spam? If there's a viral page with a share button, everyone's going to tweet nearly same text.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.