Kitz Forum
Chat => Chit Chat => Topic started by: Weaver on June 17, 2021, 11:55:37 PM
-
A friend of mine just attempted to post something on BookFace. She wrote: "I have the keyboard skills of a gnat". Facebook produced an "anti-bullying" report and deleted the post! Was she being derogatory to gnats?
-
technically, yes she was.
-
Isn't a gnat like a flea?
I got a 12 hour suspension from tweeter the other day for telling this guy who was threatening to punch someone to punch himself instead. The suspension reason was that I was supporting self harm.
-
One of my friends had something similar about a week ago saying something about sounding like a wailing cat that was also supposedly bullying.
I suspect they are automated bots picking up on key words. I've seen several instances of blocked posts over the past month or so that must be blocked by bots. I even had a post deleted of mine that was about Easter Cactus plants and then a link to a genuine wiki site... that got blocked as spam. I'm like what? :-\
-
They would need a neural network recogniser engine, crude algorithms will just produce the ridiculous results that we are seeing. And that would be expensive to scale up to their needs.
And the recognisers can so easily be defeated; just post yn Gymraeg.
-
Google Translate (https://translate.google.com/?sl=auto&tl=en&text=yn%20Gymraeg&op=translate) returns "in Welsh". Is it correct?
-
It is. My Welsh is terrible, but I’m putting in the hours in my citylit.ac.uk classes. That was where I started learning Scottish Gaelic 25 years ago before I moved to Skye to become a full-time student at smo.uhi.ac.uk.
-
Trouble is there is huge political pressure to do something about problematic posts on social media (and so there should be), but most don't understand the complexity of checking every single post, it simply can't be done by humans, you'd need huge numbers of people, so it has to be done by bots, which clearly isn't working.
-
I've just this minute had one of my Facebook posts deleted because it contains a link to a spam website.
The link was to a gov website. - More accurately it was to here: https://www.food.gov.uk/news-alerts/alert/fsa-prin-36-2021?
I dunno what algorithm they are using to think a .gov.uk site is spam :-X
-
Trouble is there is huge political pressure to do something about problematic posts on social media (and so there should be), but most don't understand the complexity of checking every single post, it simply can't be done by humans, you'd need huge numbers of people, so it has to be done by bots, which clearly isn't working.
a more sensible solution might be to apply the algorithms before the post is published, ie when you hit "send" it checks and comes back with saying you can't post it, allowing you to modify your post to make it compliant.
it will still have false positives, but it'll stop all the true positives (unless they are clever enough with the language, and those that are likely to post stuff like that won't have the language skills to by-pass it).
-
I'd imagine there are a lot more layers before it decides to even look at the post. Just doing it in real-time for every post would not always be practical.
-
I got hit by the spam bot today for something I posted AGES ago.
Apparently posting a link to https://www.dailyecho.co.uk/news/15071421.nhs-gp-practice-sets-up-a-private-service-by-paying-up-to-145-to-jump-the-queue/ is SPAM, in a discussion about the NHS being privatised! :-\