Vuukle Partners with Google to Allow Publishers to Fight Back Trolls and Haters: Perspective API Comments Plugin
The comments section has been the dark underbelly of the internet. It is not for the faint of heart. Several publishers, social media and even video gaming platform have unsuccessfully tried to fight against trolls and haters. Moderating by humans is no joy, removing comments was accepting defeat, using Facebook plugin backfired. Humanity had almost given up. Today, Vuukle announces partnership with Google to provide online publishers with World’s first Perspective API Comments Plugin. This is the most critical feature update this year in Vuukle audience engagement system.
The Jokers, the Trolls, the Spammers and the Haters
You know who they are. You have seen them. Perhaps, you have had your fingers burnt trying to reason with them. Earlier, they used to lurk in the Internet’s darkest alleys behind a veil of anonymity but now you have them shaking their tail right in your face. From the social media to publishers’ news websites, they spew venom and disrupt discussion. Psychologists have tried to explain the trolls by alluding to what they call the “disinhibition effect”. Basically, a combination of anonymity, invisibility, lack of authority, asynchronicity and isolation available online results in abandoning inhibition that we have evolved socially over millions of years. And then, you elected Donald Trump to be US President and trolling gained new heights.
Google: Helping you Fight Evil
For a company whose corporate mantra has been “Do No Evil”, Google has always had their hands filled fighting bad guys. If you are old enough to have seen the internet in the mid and late 90s, you do remember the unwanted pop-ups, porn sites and low quality websites ruling the roost. It was Google with it’s search engine ranking policies that cleaned most of the internet scum. There has been criticism of Google as it acquired more power as the default gatekeeper of the internet. But few people will disagree that Google has largely been the Dark Night of the online Gotham city. We continue to expect Google to “Do The Right Thing” and, so far, we are not disappointed.
Perspective API: What is it?
Jigsaw, an offshoot of Google, have been working on methods to fight troll and hate speech for more than an year. The team’s broader agenda, called Conversations AI, is to employ machine intelligence to develop tools that will improve conversations technology. By feeding a user comment into a deep learning network, the Perspective API is able to detect insults and hate speech and provide a toxicity rating on a scale of 1 to 100. This outperforms any blacklist of keywords or human moderators for both speed and accuracy.
Vuukle and Google join Hands
Google and Vuukle are old partners. Vuukle uses Google’s advertisement network to show relevant ads and share revenue with publishers. Google is a good choice as an advertising partner because it’s ads are not disguised as “recommended stories”. Conversation AI is Google’s open source project but Jigsaw went a step ahead. By opening an API and partnering with Vuukle, the Perspective API can be adopted by online publishers without writing a single line of code or any integration hassles. The entire number crunching is done on Google’s server and Vuukle provides the integration into its hugely popular comments widget.
Artificial Intelligence and Deep Learning to the Rescue
The Jigsaw project collected millions of annotated comments in editorial discussions from Wikipedia. If you peek into the “talk” pages on Wikipedia, they are teeming with insults and racial slurs. More often than not, editors are constantly bickering over what stays and what gets deleted. Add to this millions of user comments on New York Times and other online publishers’ stories. This, along with human annotations, resulted in enough data to train deep learning neural networks.
For example, if you type “you are not polite”, the toxic rating will be 8%, while “you are an idiot of the first order” will fetch a toxic rating of 90%. All this happens in real-time in the Vuukle comments widget. So commenters can see how “toxic” their comment is even before they press the submit button. The publisher can set the threshold for toxic rating at, say, 75% above which all comments are queued to a human moderator or dropped.
Early Adopters of Vuukle Perspective API Comments Plugin
Vuukle integrated Google’s Perspective API into it’s comments plugin and provided direct access to thousands of online publishers. The user comments from these sites will be used back into the training algorithms of Perspective API to produce more accurate “toxicity” ratings. Some of the top early adopters of Perspective API Comments Plugin are NBC-2, KEYC, Daily Sun, Ozee, BGR, Asian Age and Deccan Chronicle. So head over to any of these, click on your favorite story and try being a hater. Only this time, your words will be carefully monitored and may not pass the firewall.
