Science Google Developing Troll-Detection Software

http://www.theverge.com/2017/2/23/1...ctive-software-ai-machine-learning-developers

Google’s Jigsaw unit, as part of a larger effort to battle online trolling, said earlier today that it was releasing a new tool called Perspective, software that uses machine learning to detect harassment and abuse online.

While Jigsaw and its software are not new — it’s been around in some form since 2010, though with a different name, and Jigsaw software has even been used to spot potential ISIS recruits — this new Perspective API makes it available to developers who want to use and build on top of the software. The software works by determining the “toxicity” of online comments, a scale that has been established by mining millions of comments from the web and then presenting them to panels of 10 people (humans!) at a clip to get their feedback.

Try using Perspective to determine the toxicity of your average post

http://www.perspectiveapi.com/

@Null should embed this to grade our posts based on how toxic they are

Spoiler

 

[GethN7]

Well, both "KotakuInAction" and "GamerGhazi" scored 34%

 

[Positron]

Very progressive. I'm impressed!

 

[Alec Benson Leary]

My guess is that pedophile is a 'toxic' word in that it's often used as an accusation - so the software looks at the sentence and detects the words which are attached to this one word, along with words attached to that word. So it sees 'Pedophiles', then 'are' then 'evil', and notes they're all connected - without looking at the actual order of the words or anything before 'pedophile'. Else 'Hitler did nothing wrong' wouldn't be less than 50% toxic.

But here's what I don't get: How will this app correctly determine who is allowed to make baseless accusations and namecalling, and who is not? Will you have to submit your Twitter history so it knows if you support the right stunning-and-brave people?

 

[Trollkastel]

 

[vertexwindi]

 

[QT 219]

Well, well, well.

 

[Positron]

AGEISM!!!!!

 

[I-chi]

 

[IwegalBadnik]

Old, but worth checking into.

 

[Kari Kamiya]

And the Pokémon vs Digimon war continues.

Yooooooooo

 

[Jaimas]

I entered the infamous Geordie Tait meltdown into the thing, and it broke the scale.

I then deleted a few lines and it was down to 30%.

I then proceeded to try to hammer together a sentence. I couldn't get it past 98%.

This algorithm makes no sense.

 

[Trollkastel]

Good job Google. Amazing job.

 

[LulzKiller]

 

[ToroidalBoat]

It can't into sarcasm.

Spoiler

Also stuff written in non-Latin writing systems seems to always get a 34% score.

 

[LulzKiller]

 

[Jaimas]

Got a good one.

So close..!

 

[WireSponge]

Found another 100% phrase.

Fuck was the most toxic word I've found yet (98%) and it didn't work with only 2 so I'm beginning to think getting 100% requires at least 3 words in your text.

 

[Urban Superstition]

Hold the fuck up, Zoe Quinn and Anita Sarkeesian were involved in the making of this? God damn. I thought the use of the word "toxic" was retarded enough.
I was thoroughly convinced that SJWs had very little influence outside their little circles, but now I'm just disappointed that other people might actually be buying into their bullcrap.

 

[Sailor_Jupiter]

I should've known SJWs would be triggered by math.

"Gas all Canadians" only got the predictable "34%." Canadians are polite even when being threatened with mass-murder. ^^;;;;

 

[DICKPICSRUS]

Spoiler: Quran (2:228)

[/

]

Spoiler: Quran (24:31)

[/

]

Spoiler: Quran (4:24) and Quran (33:50)

[/

]