I’ve been getting a number of, what I presume to be, Japanese spam comments. All have been attempting to promote websites ending in uni.cc.
It also seems that the comments are being manually entered, as my first line of defence catches the comments, but the user has then pressed the manual over-ride switch to prove they are human. That, or they are using software that’s able to go through the manual verification procedure too.
Askimet doesn’t seem to catch the comments either, which is odd considering the number of links left in the comments. I wonder if there is some attempted skewing of the system going on there, through malicious users marking such comments as “not spam”.
Either way, I’ve removed the manual verification option, so that such comments can’t be posted without at least being edited first. What I really need is a way to detect input using different alphabets, so that they can be excluded automatically by the TanTan spam filter.
And hopefully Askimet will soon adapt to this type of spam enough to start catching some of it…