Advice: would expanding the pediaadvise.com website blacklist to something like !!/blacklist-website \bpediaadvise\b be a good or bad idea? Would that be a misuse of the website blacklist?
Hello @tripleee (since your icon indicates that you're here)! Has there been any attempt at using smokedetector for classifying certain text that someone already has ?
@anki I don't understand your question, Smoke Detector is very much specialized for Stack Exchange but the classification component could be run on other posts in HTML if you wanted to
I don't think it has been done successfully; my experience with other components (like Spam Assassin) is that people blindly try to apply it to whatever domain without trying to adapt it to the new domain, and are disappointed and give up when it doesn't immediately bring stellar results
@tripleee I have a suspect's profile URL, I'll get the public profile data using some tool and want to use SD to see if that profile data is spam or not.
usually a better way to investigate would be to google for the same text and see if it occurs verbatim in lots of other places or particularly suspicious places
@anki we are very good at finding recurring promotions but not particularly good at identifying generic spam
@anki for that to work we would have to have already seen the same URL before, and marked it as a spam link
still not a very good heuristic then; we have e.g. forms.gle in the patterns because we see a fair amount of spam abusing that service; that doesn't mean that Google Forms is spam
probably a better tool would be using a service with URL reputation, though I don't know off-hand of any free malware identification services like that
in my limited experience, they have the same basic flaw; URLs are fine until somebody fetches the resource and analyzes it
but someone who does this proactively for fresh links should at least have a good chance of identifying malware which is being promoted actively
in this example we could add those domains to our list, and then Smokey would indeed classify them as suspicious; but you would have to provide us with a list of domains we should add
for what it's worth a5office.co.uk resolves to an IP address which we have seen in promotions a few times
different IP address ranges have different hosting patterns but by this brief examination, I would assume that the same spammer sits behind your problem as well as the spams we see promoting these other related web sites
(for example, if this was an IP address at some of the Amazon IP ranges, you would see a bewildering mix of legitimate and spam sites, and not really be able to infer anything)
Stack generally doesn't generally care about spam profiles so we don't target those at all, but of course your particular incident might help us see a bigger picture and possibly add some checks which benefit you as well as us; but this is highly speculative
we will be happy to help you investigate, but I don't think Smoke Detector is going to help you much at all
re: IP address ranges I am afraid I'm the only one here who is currently active in that area; but feel free to ping me if you think I could help
@tripleee I'd rather clarify some things. "Headless chatbot that detects spam and posts links to it to chatrooms for quick deletion." indicates that it receives some text input, it matches the text against some database, classifies the text/username etc as spam and then tells an admin about it.
user12* was doing some machine learning on spam which might be worth checking out; Smokey currently does not perform any machine learning, so it doesn't really learn except when we painstakingly teach it, one fact at a time
have a look at findspam.py to see our rules and watched_keywords.txt for a 25k list of individual regular expressions which currently raise an alert for further investigation
@anki not a case I'm familiar with, but requiring manual user registration by email or etc seems like a small price to pay for keeping your bug tracker clean, if a CAPTCHA doesn't trivially solve it
@anki ^ that's Ryan teaching Smokey one new (rather ephemeral) fact
if you are good with Python I would not rule out the possibility that we could work out some useful synergies but nothing like a quick fix, more like if you want to commit to spending a few months with us, the result could benefit us both ... mmmmaybe
@anki something like "hello I am so and so, I would like access to your bug tracker" and then the admin sets them up with a temporary password
it's not like this is immune to abuse by any means but I would optimistically think that spammers would not even try
[ SmokeDetector | MS ] Bad ip for hostname in answer, bad keyword in answer, blacklisted website in answer, link at beginning of answer, potentially bad ip for hostname in answer, +2 more (314): Changing layer order in GIMP by sanemine on graphicdesign.SE
@tchrist That message is actually forwarded from metasmoke. metasmoke posts that message in here the first time you give feedback via chat on each chat server, so up to three times. It will be in here for feedback that was given on chat.SO or chat.meta.SE, which can be a bit confusing. It does not take into account any feedback which was done via the MS API or on MS itself.
I just received this review audit in the Low Quality Posts queue. Here is a screenshot:
I passed the audit, but only because I noticed that the user who posted it was deleted, so I opened the post in a new tab to see if it was an audit. If the user wasn't deleted, I would probably have pressed L...
@Mast Agreed. I was writing up an answer (not including links to MS so that the spammer doesn't catch on, if they look that way) and I wanted to see how long ago the trend started.
Hello you all. stackoverflow.com rescued me countless times and now I want to do something useful for it too. I began answering questions and want to help with Smokedetector too. Some people in SObtics chat said me I need to go here and ask people there about it. So I ask.
@VictorVosMottor I am aware of 8 admins. Currently in this room: Makyen, Undo, thesecretmaster, ArtOfCode. Not currently in this room: Thomas Ward, Andy, angussidney, tripleee.
@YeeHaw I suggest you read the Introduction to Participating in Charcoal. I can not create an account for you (not won't, can't). You would need to create the account for yourself. OTOH, I can turn on the general ability to create accounts. The restriction that is preventing you from creating an account isn't you; it's everyone. The reason the restriction is in place at the moment is that there's a troll who has been creating accounts.
However, even once you create an account, without more participation on Stack Exchange sites, I'm not going to provide additional privileges. Unfortunately, the spammers and trolls do watch this chat and metasmoke. They do try to participate here to gain privileges. As a result, we require that there has been more participation on Stack Exchange than you currently have.
However, you can help flag reports on Stack Overflow. Once you have 200 reputation there, or on any site, you will be able to get an "association bonus" on all other sites, which will allow you to flag on other sites. Keep in mind when looking at reports that just because it was reported here doesn't mean it's spam or R/A. You must use your own judgement. You need to keep that firmly in mind for Stack Overflow in particular, because it has the highest rate of false positives of all sites.
