Board index › Topical Discussion › Computers, Math, Science, and Technology

troll detection algorithms

Page 1 of 2 [ 22 posts ]  Go to page 1, 2  Next

Previous topic | Next topic

anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

20 Jul 2016, 6:10 pm

does anyone know of (or has anyone contributed to) anything related to automated troll detection? it's an idea that had never occurred to me before, but it makes a lot of sense, it's obviously something increasingly relevant, and it's the kind of thing that would involve most of my main interests all at once (computational linguistics, data mining, psychology, social dynamics, among others)

any tips on where to start if i'm looking for relevant studies/data and existing projects?

thanks!


_________________
404


Ichinin
Veteran
Veteran

User avatar

Joined: 3 Apr 2009
Gender: Male
Posts: 3,653
Location: A cold place with lots of blondes.

21 Jul 2016, 2:33 am

https://cs.umd.edu/~srijan/pubs/trolls-asonam14.pdf

https://kth.diva-portal.org/smash/get/d ... TEXT02.pdf

https://www.technologyreview.com/s/5366 ... ial-trade/


_________________
"It is far better to grasp the Universe as it really is than to persist in delusion, however satisfying and reassuring" (Carl Sagan)


anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

21 Jul 2016, 3:04 am

Ichinin wrote:
https://cs.umd.edu/~srijan/pubs/trolls-asonam14.pdf

https://kth.diva-portal.org/smash/get/d ... TEXT02.pdf

https://www.technologyreview.com/s/5366 ... ial-trade/

nice. thanks! :)


_________________
404


naturalplastic
Veteran
Veteran

User avatar

Joined: 26 Aug 2010
Age: 70
Gender: Male
Posts: 35,189
Location: temperate zone

21 Jul 2016, 9:36 am

"Troll farms"?

That's a new thing I never heard of before.



anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

21 Jul 2016, 8:08 pm

naturalplastic wrote:
"Troll farms"?

That's a new thing I never heard of before.

i hadn't heard of it either. sad thing. so the cat-and-mouse game has begun...


_________________
404


Jono
Veteran
Veteran

User avatar

Joined: 10 Jul 2008
Age: 44
Gender: Male
Posts: 5,660
Location: Johannesburg, South Africa

22 Jul 2016, 1:01 pm

One problem that might occur with an such an algorithm is that there could be false positives for detecting trolls. In other words, there's a possibility that an ordinary member who is not anti-social but and not a troll might be identified incorrectly as a troll by the algorithm and needlessly banned and that's especially a problem if there are no human moderators to look over their post history and use their discretion to decide if they're really trolls or not.



anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

22 Jul 2016, 2:29 pm

Jono wrote:
One problem that might occur with an such an algorithm is that there could be false positives for detecting trolls. In other words, there's a possibility that an ordinary member who is not anti-social but and not a troll might be identified incorrectly as a troll by the algorithm and needlessly banned and that's especially a problem if there are no human moderators to look over their post history and use their discretion to decide if they're really trolls or not.

oh yes, there always have to be human moderators around. even assuming a detection system could be virtually perfect at one point in time (which is not possible), it could never stay perfect on its own without direct human input. and who wants robots to be telling people what's right or wrong anyway, right?

the idea (at least the idea that i have in mind) is to make the work easier for moderators. it's hugely easier to browse a list of probable trolling attempts (with explanations about why they've been automatically flagged) than browse all posts, or rely on forum members to report all occurrences (it also gets tiresome for forum members when there's too much to be reported). then moderators can focus on the human aspect instead, making posts of their own to set the tone of the interaction


_________________
404


somebody300
Blue Jay
Blue Jay

User avatar

Joined: 8 May 2016
Age: 29
Gender: Male
Posts: 93
Location: France

23 Jul 2016, 6:05 pm

We have so many useful problems in computer science that would lead to much benefit to the society if they could be solved, and so many algorithms which could be developed. Ranging from those which pertain economic accounting and statistics to those which could be used in exact sciences.

But then some people think that we need troll detection algorithms. Algorithms which would require thousands of hours of interdisciplinary expert work, and cost hundreds of thousands of dollars to develop.
I don't really understand this kind of thinking. I mean, what could lead someone to post such an idea? Or is it some kind of subtle sarcasm that I don't understand?



anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

23 Jul 2016, 6:47 pm

well... ever heard of a hobby?

besides, i'm not really concerned with how useful my hobby could be for society, other than for its potential profitability. and even that is not a decisive factor anyway. all i'm really concerned with is that it would be fun :D. oh and i don't think it's going to cost a dime. i can do it myself in my spare time. i'm that smart. and that humble :mrgreen:

either way, even assuming it's not something very useful, by my standards it's already way less redundant than the work i actually get paid to do

and... if you don't think that intentionally and virally disruptive behavior in anonymous environments is a relevant issue for society... think again. read the news. think trump. think erdogan. think putin. and so on. and their opposition as well. they know what i'm talking about. and if you don't see that, then you'll get taken advantage of


https://www.youtube.com/watch?v=rE3j_RHkqJc


_________________
404


somebody300
Blue Jay
Blue Jay

User avatar

Joined: 8 May 2016
Age: 29
Gender: Male
Posts: 93
Location: France

24 Jul 2016, 12:51 pm

anagram wrote:
well... ever heard of a hobby?

besides, i'm not really concerned with how useful my hobby could be for society, other than for its potential profitability. and even that is not a decisive factor anyway. all i'm really concerned with is that it would be fun :D. oh and i don't think it's going to cost a dime. i can do it myself in my spare time. i'm that smart. and that humble :mrgreen:

either way, even assuming it's not something very useful, by my standards it's already way less redundant than the work i actually get paid to do

and... if you don't think that intentionally and virally disruptive behavior in anonymous environments is a relevant issue for society... think again. read the news. think trump. think erdogan. think putin. and so on. and their opposition as well. they know what i'm talking about. and if you don't see that, then you'll get taken advantage of


https://www.youtube.com/watch?v=rE3j_RHkqJc


Well yes, if you mean paid shills/trolls (the Communist Party of China and Putin do pay people to shill and disrupt online communities opposing their views), then this could be useful. But it still would demand a lot of resources.
Why not to just use some current and already developed anti-spam technologies (which use ANNs and can learn trolling patterns of behaviour), and apply them to that?
Moreover, they could be integrated to work with a webcrawler piece of soft, and effectively detect online shilling by the CPC/Putin/ISIS/etc.

...
Any Artificial Neural Network, trained in a way as not to result in overfitting, can do this kind of job. ANNs can detect patterns of behaviour, and even perform at the level of a human (or even better) in some tasks of pattern recognition and concept formation.



anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

24 Jul 2016, 2:53 pm

doubly missing the point again. first: it's a hobby. it makes no difference if it's useless. "why not just <any other idea> instead?". simple. because i don't care about <whatever other idea you have in mind>. second: that was not what i was talking about. i don't see a point in trying any further to explain it though, when it's clear that you're only interested in recursive cherry-picking of arguments with no regard for context

bam! troll pattern detected :)


_________________
404


vermontsavant
Veteran
Veteran

Joined: 7 Dec 2010
Gender: Male
Posts: 6,110
Location: Left WP forever

24 Jul 2016, 3:01 pm

it would take some time to indentify trolls through there system,by the time there discovered the damage is done.remember a troll only needs one post to do damage


_________________
Forever gone
Sorry I ever joined


somebody300
Blue Jay
Blue Jay

User avatar

Joined: 8 May 2016
Age: 29
Gender: Male
Posts: 93
Location: France

24 Jul 2016, 3:17 pm

anagram wrote:
doubly missing the point again. first: it's a hobby. it makes no difference if it's useless. "why not just <any other idea> instead?". simple. because i don't care about <whatever other idea you have in mind>. second: that was not what i was talking about. i don't see a point in trying any further to explain it though, when it's clear that you're only interested in recursive cherry-picking of arguments with no regard for context

bam! troll pattern detected :)


If it's a hobby, you would need to be familiar with a large amount of topics in advanced computer science, mathematics and statistics to accomplish this, and probably have hundreds of thousands of dollars to spend. It isn't something which can be done by someone in their bedroom, for example.



anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

24 Jul 2016, 3:21 pm

vermontsavant wrote:
it would take some time to indentify trolls through there system,by the time there discovered the damage is done.remember a troll only needs one post to do damage

that's true, and it's one of the reasons why such a system is inherently limited. one possible approach to that specific problem though (posts that are particularly inflammatory on their own) is preemptive flagging, calling for a moderator to look at it before it's made public

btw, i do see how all of these concepts can be used for systematic censorship of any nature and for any intent. that's something to be taken into account at all times


_________________
404


anagram
Veteran
Veteran

User avatar

Joined: 4 Nov 2012
Gender: Male
Posts: 1,433
Location: 4 Nov 2012

24 Jul 2016, 3:30 pm

somebody300 wrote:
If it's a hobby, you would need to be familiar with a large amount of topics in advanced computer science, mathematics and statistics to accomplish this

what a fortunate coincidence then :)

you didn't stop to think that you don't know about my individual skills and background, did ya?

Quote:
and probably have hundreds of thousands of dollars to spend.

funded projects aren't the only thing in the universe. certain things don't depend on material resources. time is the main resource involved here. also, the end result is not the main goal. development itself is what matters. it's a hobby

Quote:
It isn't something which can be done by someone in their bedroom, for example.

i've been told similar things in the past. didn't stop me, and i proved them wrong

besides, ever heard of linus torvalds and the linux system? or steve wozniak and the personal computer?

you're pretty lousy at raining on parades, you know? :)


_________________
404


somebody300
Blue Jay
Blue Jay

User avatar

Joined: 8 May 2016
Age: 29
Gender: Male
Posts: 93
Location: France

24 Jul 2016, 3:57 pm

anagram wrote:
somebody300 wrote:
If it's a hobby, you would need to be familiar with a large amount of topics in advanced computer science, mathematics and statistics to accomplish this

what a fortunate coincidence then :)

you didn't stop to think that you don't know about my individual skills and background, did ya?

Quote:
and probably have hundreds of thousands of dollars to spend.

funded projects aren't the only thing in the universe. certain things don't depend on material resources. time is the main resource involved here. also, the end result is not the main goal. development itself is what matters. it's a hobby

Quote:
It isn't something which can be done by someone in their bedroom, for example.

i've been told similar things in the past. didn't stop me, and i proved them wrong

besides, ever heard of linus torvalds and the linux system? or steve wozniak and the personal computer?

you're pretty lousy at raining on parades, you know? :)


The fact that you can't do a basic literature review (something which is a part of any basic scientific curriculum), and have to resort to asking people on an autism forum to do it for you makes one think that you aren't exactly the next Linus.