As part of a bigger effort to prevent online trolling, Google builds an app that uses learning mechanisms and can detect abuse and harassment on social networks.
Google’ Jigsaw unit wants to implement an app called “Perspective”. The purpose of the app is to prevent and remove social networking comments that can affect users. Posts are being analyzed by analyzing millions of web comments by presenting them to some groups of ten people. The idea behind this is to gather people from multiple social categories and learning about their opinions about online trolling.
A demo of this app can be found on the Perspective API website and anyone has the possibility of using it to see how abusive or toxic his or hers comments are.
Although the app created by Jigsaw is not new, the app is now accessible and it’s also used by quite a decent number of companies. Publications such as The New York Times, The Guardian and The Economist are currently experimenting with this tool and using it in the comments section of their website.
From what we know so far, some phrases that are being used more and more lately have a pretty high percentage of being tagged by Perspective. For example, the “fake news” expression has a level of “toxicity” of 47% , “Bad hombre” gets 55% and “Grab her by the pussy” earns 92%.
Because Googles app uses similar comments, which were tagged as inappropriate, it is pretty limited and cannot identify specific nuances in the phrases it analyzes. However, considering the fact that the app is based on a learning mechanism, we think it will become smarter in the future.