@bmike You left a comment on the post, but declined to delete it. I assume you're reporting it here so we have a record of it in case more posts are made about the application/domain. Is that correct?
@Makyen Welcome to Charcoal HQ @Mr.YouKnowWhoIAm! I'm SmokeDetector, a bot that detects spam and offensive posts on the network, and posts alerts to chat. You can find more about me on the Charcoal website.
@Mr.YouKnowWhoIAm Regular users can sign up for a metasmoke account, as you've already done, and participate in autoflagging on the sites where they have enough reputation to cast flags. They can also participate by giving feedback to SmokeDetector and manually flagging posts which are spam or rude/abusive. To give feedback to SmokeDetector, you will need to be granted that authorization.
@Makyen I am not going to get 300 rep on any of the 174 SE sites (if that is accurate) any time soon. Is there a minimum amount of rep to do something?
@Mr.YouKnowWhoIAm As long as you have 15 reputation, you can raise flags. You could watch for reports then go to the site on which the post is located and flag the post, if it is spam or rude/abusive. You could also participate in autoflagging. If you want to do coding, you could look through the issues in GitHub.
I, unfortunately, have to go now. I'm being called away in real life.
@Mr.YouKnowWhoIAm I am aware of 7 admins. Currently in this room: Makyen, thesecretmaster. Not currently in this room: angussidney, tripleee, ArtOfCode, Andy, Undo.
@Mr.YouKnowWhoIAm I'm relatively new to being an admin, so I'm not really sure exactly what you're asking for. I'm assuming you mean on metasmoke. Is that correct? If so, what exactly is it saying?
@Mr.YouKnowWhoIAm Are you talking about autoflagging?
@Mr.YouKnowWhoIAm If you were asking about autoflagging on metasmoke, I have given you the "Flagger" role, which should permit you to participate in autoflagging.
@Mr.YouKnowWhoIAm Because it's helpful to know that information in some cases. For example, there are multiple instances of SmokeDetector running and when switching form one to another we may want to be running an older version (i.e. there might be a bug in a newer version), or we may need to know that the instance needs to be updated to the current revision.
@Mr.YouKnowWhoIAm Metasmoke holds the database of posts. It presents that data in ways that's easy for users to consume. It also provides that data to the various userscripts we have. In addition, metasmoke is also in charge of autoflagging.
@Mr.YouKnowWhoIAm That depends on what you want to do. Most of the information is accessible through the various menus and links. Without more to go on, it's difficult for me to say much more than that you should explore the site and see what you find interesting.
@Makyen What should my password be on metasmoke by default? I cannot find my password. I signed in using my SE account. I already tried my SE password.
@Mr.YouKnowWhoIAm If you mean from a personal standpoint, then there is a risk of casting a flag which gets declined. How much of a risk that is depends on your choice for autoflagging conditions, which you can set on metasmoke. When setting conditions, you will be shown how many of the posts in the database would have been flagged and what percentage of those were "true positive" (i.e. that were marked as spam by people reviewing).
@Mr.YouKnowWhoIAm You should be able to log back in using Stack Exchange. The last information I heard was that metasmoke accounts are rarely associated with a password that anyone knows (i.e. it's randomly generated and not kept), unless you created the metasmoke account as not associated with your Stack Exchange account.
@Mr.YouKnowWhoIAm Yes, but that is done through Stack Exchange's OAuth2 flow. When you begin the authentication from metasmoke, it should navigate to a Stack Exchange URL which then will ask you to confirm. You may need to enter your Stack Exchange credentials (into the Stack Exchange site, not metasmoke) at that time.
@Makyen I already tried to write-authenticate (the first time the SE link opened), but it keeps on saying that I have to write-auth inside “Clifford the Big Red Button” (nice name, by the way).
@Mr.YouKnowWhoIAm Ah... that. My understanding is that's a bug. I believe it displays that even if you are already authenticated (e.g. it displays that for me too).
@Mr.YouKnowWhoIAm Yes, the majority of commits are adding or removing entries on our lists of regexes or numbers which each post is checked against. A brief description of them can be found here in the commands.
@Shree Looks like it's another instance where not having the website blacklist bookended has resulted in an entry catching a site which is not intended.
@Mr.YouKnowWhoIAm Personally, I have those set at 100, the maximum, on all sites except the main site I use, Stack Overflow. On that site, I have it set slightly lower. However, the reality is that it is very rare for autoflagging to use a large number of flags on any site. To the best of my knowledge, it's only happened once. It was on a site and on posts which only a limited number of people had autoflagging conditions set such that the posts qualified and there was our largest wave of spam.
On that occasion, about a month ago, every single flag for all people who had set up appropriate conditions was used. By the time the spam wave was over (it lasted about an hour), metasmoke had run out of available flags on that site for posts that fit those conditions.
@Mr.YouKnowWhoIAm Looks OK, but just so you know, Stack Exchange has a maximum of 100 flags on posts plus 100 flags on comments per UTC day for regular users.
@Machavity If you're looking for a delimited regex for the website blacklist, so far I've been using a construction like \b(?<!-)binfer\.com(?![.-])\b. If you have other ideas for delimiting it, I'm interested in what might be better.
@Mr.YouKnowWhoIAm There is no set minimum for weight. What is enforced is that your autoflagging condition must have an accuracy of at least 99.75%. When setting the condition you can click "Preview" to see what the accuracy is.
@Mr.YouKnowWhoIAm Ah, I misunderstood. The "Minimum Weight" is the minimum weight which the post must have in order to qualify for your condition. That, of course, then begs the question "what is weight?" Each detection reason we have has a "weight" associated with it. That weight is the equivalent of it's accuracy percentage based on the historical data. So, a weight of 75 for a detection means that 75% of the time that detection is triggered, the post was marked as "true positive" by reviewers.
For a few detections the % accuracy isn't reflected in their weight. If the detection is considered "experimental", then it's weight is locked at 1. If the detection has detected too few posts, then it's weight is held at 0 until it has detected enough posts such that the weight is at least semi-meaningful.
The "weight" for a post is the sum of the weights for the detections which were triggered for that post. It is the post's weight that is being compared with the number you select for "Minimum Weight" in your autoflagging conditions. The post's weight must be equal to or greater than that "Minimum Weight".
@Makyen So basically, the minimum weight is the minimum probability of the accuracy of the detection that a post has to be to be autoflagged. Am I right?
What is it with binfer.com that requires you to blacklist it?
@Machavity Yeah, it shouldn't be left off. It would be a good idea to change the keyword blacklist to binfer(?!\.com) instead of binfer, so that it's not also detecting the domain. Unless there is a specific intent to so do (and I'd expect someone to leave a (?#comment) if they did), then we normally don't want two detections triggering off the exact same thing. The blacklisted keyword should detect mention of the product "Use Binfer. It's ...", but shouldn't detect the domain.
@Mr.YouKnowWhoIAm It gets used in spam on a regular basis. Often the posts for binfer are difficult to determine that they are spam, as they can look like very low quality posts, if they are answers placed on related questions.
@SmokeDetector That's a "Potentially bad keyword", which means it's on the watchlist, which is an experimental reason and has a weight of 1. That particular watch is a watch for posts that begin with a link, excluding various domains, and excluding posts that contain <code> or <blockquote>. It's on line 7058 in watched_keywords.txt. There is also one above it on line 7057 which is only for YouTube.
Eventually, the intent is that it will be it's own detection. With specific domains excluded, it does a reasonable job of finding spam that we don't otherwise detect (e.g. from domains we haven't previously seen). The down side is that it does have a significant number of false positives.
The last I checked, which was a while ago now, it was at about 59% TP (true positive) over the posts its detected since it was added.
[ SmokeDetector | MS ] Url in title, bad keyword in body, bad keyword in title, body starts with title and ends in url, potentially bad ns for domain in body, +3 more (392): downthepch.com/level8-keto/ by xdma2ch on wordpress.SE
@Machavity I think you may have not been taking into account the bookending that's done on the watchlist and blacklisted keywords. The regex which you have used is equivalent to this search on MS, which detects only 29 of the 60 TP which are caught by binfer or binfer(?!\.com), which only detect 1 FP from > 1 year ago & that one is debatable.
I'm not sure what you're trying to exclude or include by having the .*(transfer|send|sharing|alternative|service|files).
For the watchlist and blacklisted keywords, all of the entries are concatenated together, separated by | and surrounded by (?is)(?:^|\b|(?w:\b))(?: before the entries and )(?:\b|(?w:\b)|$) after the entries. In most cases, this means that you should be doing searches with the regex surrounded by \b. In addition, when using . you need to set the s bit in the search. I commonly do \b(?s:my regex)\b, but there are multiple ways to get the equivalent search (e.g. (?s)\bmy regex\b).
In addition, for performance reasons, it's very helpful to limit constructions like .* to a maximum number of characters (e.g. .{0,400}) rather than an unlimited number. This can help significantly in reducing the amount of computing power needed, particularly when checking long post bodies that don't match. Sometimes this is not possible, but we try to use it where .* isn't needed.
For searches where you think the next thing will be found close to the earlier part of the regex, it's often a good idea to use ? to cause the regex engine to search forward rather than go to the maximum extent and backtrack. This is also very helpful if you think the tail of the regex might match multiple times, as not using ? will result in only a single match for the entire post, rather than multiple matches. That also can result in a very long string included in the why data.
However, it does have some side effects, which can be an issue under some relatively unusual conditions.
BTW: @ArtOfCode Are you sure this post is FP? It looks quite similar to the rest of the Binfer posts.
Sorry, never mind. It's on Software Rec and is basically requested, so even though it's probably by the same people, it's not actually spam.
OTOH, it's not actually a cloud service, so isn't actually what the question is asking for.
@Makyen The binfer site also doesn't appear to mention an ability similar to what the OP calls "binary diff sync", which is one of the two requirements in the question. Having now looked into Binfer a bit more, it doesn't appear to meet either of the two requirements which are asked for in that question.
@Makyen That pattern looks like it's already caught by Pattern-matching website in answer and Pattern-matching website in body; append -force if you really want to do that.
@Machavity It looks like you're gone for the day. I'm going to change the blacklisted keyword to binfer(?!\.com). You're welcome to work on it more later.
[ SmokeDetector | MS ] Blacklisted website in body, link at end of body, pattern-matching website in body, potentially bad ns for domain in body, potentially bad keyword in body, +1 more (298): Where to buy LaPura Cream by perturner on meta.SE
[ SmokeDetector | MS ] Url in title, bad keyword in body, bad keyword in title, body starts with title and ends in url, link at end of body, +5 more (591): downthepch.com/level8-keto/ by besbcob on meta.SE
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body, body starts with title and ends in url, link at end of body, pattern-matching product name in body, +4 more (689): You can choose to brew inexperienced by user104162 on workplace.SE
[ 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, +2 more (685): Ketoviante Philippines by wazmjkoplod on askubuntu.com
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, blacklisted website in body, pattern-matching website in body, potentially bad ip for hostname in body, +2 more (492): Keto Buzz Tablets by MishelConz on wordpress.SE
I did notice those other two answers, but at a cursory glance they didn't seem any worse than NAA. Reading them properly now, it seems that I was wrong :) Good pick up
Hey there, can someone who knows Hindi look at this? Google translate looks like there might be some bad content in there but I don't want to report because I am not sure.
[ SmokeDetector | MS ] Url-only title, bad keyword in body, blacklisted website in body, blacklisted website in title, body starts with title and ends in url, +5 more (791): livesmartt.com/keto-flex-diet/ by mustna12 on askubuntu.com
@SmokeDetector @AJ we were wondering if it would be possible for you to confirm that this answer would be classed as R/A (the google translate result suggests so, but we're not entirely sure)
@SmokeDetector My problem with thinking this is spam is that the site would almost never want people to think that there was a problem with getting their money out. Such a site relies on people being gullible enough to think that they are actually going to double their money in 24 hours.
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body, potentially bad ns for domain in body, potentially bad keyword in body (197): S-Nail Mail Error by Alkatone69 on askubuntu.com
@SmokeDetector in the absence of any evidence to prove that they are affiliated, and given the fact that the link does seem to answer the question, I'm leaving this as NAA
[ SmokeDetector | MS ] Url in title, bad keyword in title, blacklisted website in title, pattern-matching product name in body, potentially bad ip for hostname in title, +3 more (495): getmagnetiques.com/fantastic-keto/ by Crues mtaxer on academia.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, +8 more (888): supplementeffects.com/vanitrol-uk/ by rkdfkzrpd on wordpress.SE
[ SmokeDetector | MS ] Bad keyword in body, pattern-matching website in body, potentially bad ns for domain in body, potentially bad keyword in body (195): Where To Order It? by donna spivey on meta.SE
@SmokeDetector @angussidney when this was previously reported (5hrs ago), you gave tpu- feedback and now fp-. Can you explain this and possibly retract one feedback?
@SmokeDetector the revision of the post contains a link which (based on the username) the user is affiliated with, but did not disclose, which is why I cast a spam flag and marked as tpu-. The subsequent revision by Glorfindel has edited out the link, hence making the second revision fp-.
@SmokeDetector the previous revision of the post contained a link which (based on the username) the user was affiliated with, but did not disclose, which is why I cast a spam flag and marked as tpu-. This revision, edited by Glorfindel, has edited out the link, hence making this second revision fp-.
IIRC our policy on this is to provide feedback on each post in MS as per the specific revision stored in MS (regardless of whether previous/subsequent revisions are spam), but if I'm wrong feel free to correct me
[ SmokeDetector | MS ] Url-only title, bad keyword in body, bad keyword in title, body starts with title and ends in url, link at end of body, +4 more (494): downthepch.com/level8-keto/ by noori4 on askubuntu.com
[ SmokeDetector | MS ] Url-only title, bad keyword in title, blacklisted website in title, potentially bad ns for domain in title, potentially bad keyword in title (299): awaretalks.com/level8-keto/ by jameywarsish on askubuntu.com
@SmokeDetector SmokeDetector's flag on this post was blocked because of a ban, but it looks like it has since lifted because another post was successfully flagged (suggesting that either the bug on SE's was fixed, or we didn't run into it this time)
[ SmokeDetector | MS ] Url-only title, bad keyword in body, blacklisted website in body, blacklisted website in title, body starts with title and ends in url, +6 more (891): livesmartt.com/keto-flex-diet/ by werin201 on askubuntu.com
[ 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, +5 more (788): livesmartt.com/keto-flex-diet/ by ytfdma12 on workplace.SE
[ SmokeDetector | MS ] Url in title, bad keyword in title, blacklisted website in title, pattern-matching website in title, potentially bad ip for hostname in title, +3 more (497): mummydiet.org/marathon-keto/ by Spor jahrt on ham.SE
