[ SmokeDetector | MS ] Blacklisted website in answer, potentially bad ip for hostname in answer, potentially bad ns for domain in answer (95): A/B testing software by raj111 on askubuntu.com
[ SmokeDetector | MS ] Blacklisted website in body, potentially bad ip for hostname in body, potentially bad keyword in body, potentially bad keyword in title, blacklisted user (173): Mysore One Day Trip by Best Places To Visit In Mysore on stackoverflow.com
I find that suspicious: a user comes to SO to post a question about a topic which is not even closely related to programming and gets spammed quite often
I'm with the Moose on this one. There's nothing to really ascribe that TP to other than a search for that user. Sure, it's nonsense, but it's not R/A nonsense, and it's also not necessarily spam. Suspicious? Yeah, but not enough to think of it as a "true positive."
What caught that post in the first place was an accidental connection between the end of one sentence and the next anyway.
[ SmokeDetector | MS ] Bad keyword in link text in answer, pattern-matching website in answer, potentially bad ns for domain in answer, potentially bad keyword in answer (161): Boundary of Siegel modular variety by akash gill on mathoverflow.net
@Rubiksmoose Dang it. I missed that part. I was previewing it from a (relatively small) generated image, because my sandbox is down at the moment. Thanks.
@Rubiksmoose I find myself switching gears a LOT when I don't have a filter on. It's a bit disorienting. I attribute that to the sheer number of reviews that're in the queue, though.
@Rubiksmoose In general, we prefer that people do not give feedback on posts which they report. That's not as important now, given that we require two feedbacks per post, but is still something we prefer to avoid.
@Rubiksmoose It's a legitimate service that's used to host spam
user435118
> Domains that are confirmed not spam. The fact that a domain is not spam does not mean no posts which link to it are spam, just that the purpose of the domain is such that false positives are somewhat likely.
@Daniil probably should have checked that. Thanks. Though in this case, fps are almost nonexistent. Looks like the tag update fixes that discontinuity though.
@NobodyNada I opened that up in FIRE and was like "Yikes that's a troll right" then opened up MS and saw that I had FP'd it earlier. Whoops. Changed now.
@Mast We switched, because \W* doesn't detect when spammers use _ as a separator between their words. Using a possessive quantifier (e.g. *+) is better when you know that what's following the thing you're quantifying can not be matched by what you're quantifying. Using it prevents backtracking, which know won't be beneficial to do, because the next thing can't match.
Ideally, we'd go through and change/update most of the existing regexes, but there are cases where making the change isn't appropriate (i.e. where it will cause failures to match), so it can't just be a global search and replace. I've made the change in some places when looking at a set of regexes, but I haven't felt it a high enough priority to go through all of them. There are a few/several regex adjustment tasks I'd work on prior to doing that (i.e. ones with a higher benefit/cost ratio).
[:55539637] Potentially bad keyword in username - Position 0-11: John Balvin Blacklisted user - blacklisted for //law.stackexchange.com/questions/56281 (https://metasmoke.erwaysoftware.com/post/264368) by the metasmoke API