@Stormblessed Another time, we don't report NAA posts to SmokeDetector. A) it really doesn't do any good, as a single NAA flag puts it in the review queue and multiple NAA flags do very little. B) SD's intent is to catch spam and R/A posts. Catching NAA is just a side-benefit.
@Stormblessed It leaves a record in MS as to what the post was. Currently, the record is then used to determine how accurate each detection is (i.e. "weight"). The weight of each detection on a post is then combined and it is determined if it meets the autoflagging conditions people have set on metasmoke. If it has met those criteria, then automatic flags are raised. If you use some of the userscripts, then it will raise a flag for you at the same time that feedback is given (e.g. FIRE).
@Stormblessed There is a desire to develop some machine learning from the data to add additional accuracy to detections and autoflagging. There have been a couple of starts on that, but nothing has actually been fruitful.
@Stormblessed You don't need to use a userscript, but they can make the process of reviewing posts, giving feedback and raising manual flags significantly easier. To participate in autoflagging, you have to sign up on metasmoke.
@Stormblessed There has been at least one time when we ran completely out of available flags for autoflagging on one site. There may have been other times, but I'm not aware of them. The time I'm aware of was a result of a large and rapid wave of spam where the detections were not sufficient to meet everyone's flagging conditions who could flag on the site. The site was also relatively less traveled, so those that had made flags available didn't have the full 100 to begin with.
@Stormblessed Yes and no. There's been some effort, but it depends on what you feel qualifies. Our list of bad keywords is running 96.24% TP in question bodies (over 56k detected posts) and 97.95% TP in question titles (38.7k detected posts).
@Stormblessed It was actually on a good sized site: Ask Different. However, we've only detected 7,623 TP posts, which doesn't result in everyone having all that many flags, unless they also participate in other ways.
For example, I only have about 40, or so, flags available on that site.
@Stormblessed I'm not sure which numbers you're looking at.
@Machavity I kind-of figured it had something to do with that, but I'm not sure which numbers exactly.
@Stormblessed There are currently 6,210 reviews waiting to be done. It looks like there have been 25,339 reviews done. Metasmoke has a record of: 192,340 total reports from 143 reasons: 137,075 (71.27%) TP, 45,299 (23.55%) FP, 10,513 (5.47%) NAA. There are some SD reports which metasmoke has no record of due to being down when the report came in.
@Makyen That pattern looks like it's already caught by Bad keyword in answer, Blacklisted website in answer, Bad keyword in body, and Blacklisted website in body; append -force if you really want to do that.
@Makyen That pattern looks like it's already caught by Blacklisted website in answer and Blacklisted website in body; append -force if you really want to do that.
@Makyen That pattern looks like it's already caught by Potentially bad keyword in body and Potentially bad keyword in answer; append -force if you really want to do that.
@Makyen That pattern looks like it's already caught by Potentially bad keyword in body and Potentially bad keyword in answer; append -force if you really want to do that.
@Stormblessed @Machavity If you're interested, there's more data about that spam wave in this GitHub issue. That issue is about what we can do to have SD respond better in a similar situation.
@JeffSchaller Another time, you can redo the watch. You can ping someone in here and/or leave a comment on the GitHub PR that it was superseded by another watch.
@Stormblessed That's the total number of reviews which are queued. There are 3 types of reviews: Posts, Untagged Domains, and flags for admin attention. In an ideal world, the total shown for reviews should be the total available to you. It appears that's a bug.
@Stormblessed If you mean the domain abdullah\.today, then I added it because the user who posted it did so without disclosing their affiliation. We have a record of them posting that domain on two SE sites over 4 months, the first was NAA, while the second had undisclosed affiliation. Both posts have been deleted, but it's unclear if the user was informed of the disclosure requirements.
I watched it because I consider it likely they will post again possibly on another site and potentially having changed their username, which would make the detection of "Username similar to website" ineffective. It's probably going a bit beyond what we need/should be watching.
However, that does bring up the point that I should make sure that the user is not blacklisted, as we don't want to hit three detections.
@Stormblessed The entries on all of the blacklists and the watchlist (with the exception of the two numbers lists) are regular expressions. In a regular expression a . normally matches any character. Putting a \ in front of the ., as in \. results in the . not having any special meaning and matching only an actual ..
@SmokeDetector This is the user's 3rd answer promoting the product (at least 1 deleted). They have no contributions other than promoting this product. It is a product which we are watching because it is promoted.
[ SmokeDetector | MS ] Bad keyword in body, bad pattern in url body, blacklisted website in body, pattern-matching website in body, potentially bad ns for domain in body, +2 more (487): A Hammock in the Bedroom? Why Not? by Shanu Sweet on english.SE
[ SmokeDetector | MS ] Bad keyword in body, bad keyword with email in body, pattern-matching email in body, potentially bad keyword in body (236): how i get my ex back by Cynthia Ikwueme on interpersonal.SE
[ SmokeDetector | MS ] Bad keyword in body, link at end of body, link following arrow in body, potentially bad ns for domain in body, potentially bad keyword in body (295): How to utilize Yoo Slim? by yooslimkonjac on meta.SE
[ SmokeDetector | MS ] Url in title, bad keyword in body, blacklisted website in body, blacklisted website in title, body starts with title and ends in url, +6 more (692): slimtonepills.com/biogenx/ by user141553 on graphicdesign.SE
Warning: teward/Supergate: 10 latest connections to metasmoke have failed. Disabling metasmoke
Warning: Makyen/EC2-linux: 10 latest connections to metasmoke have failed. Disabling metasmoke
[ SmokeDetector | MS ] Url in title, blacklisted website in body, blacklisted website in title, body starts with title and ends in url, pattern-matching product name in body, +5 more (596): slimtonepills.com/biogenx/ by romyneny on graphicdesign.SE
@MartinSleziak Potentially problematic ns configuration in body - Suspicious nameservers: all IP addresses for www.examfear.com are in set {'103.35.165.65'}
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, blacklisted website in body, body starts with title and ends in url, link at end of body, +1 more (587): PayPal Customer Service Number by miller logan on askubuntu.com
[ SmokeDetector | MS ] Blacklisted website in body, body starts with title and ends in url, link at end of body, pattern-matching website in body, potentially bad ns for domain in body, +2 more (396): Biogenx As referenced previously by ranjumax on askubuntu.com
[ SmokeDetector | MS ] Bad keyword in body, bad pattern in url body, blacklisted website in body, pattern-matching product name in body, potentially bad keyword in body, +1 more (389): Wonder Full Keto! Coat & Views by fifiriry on drupal.SE
@Machavity That pattern looks like it's already caught by Potentially bad keyword in answer and Potentially bad keyword in body; append -force if you really want to do that.
