It goes by many names – toxic commentary, hate speech, uncivil discourse – but no matter what you call it, research has shown that abusive and harassing online exchanges are harmful to the free flow of civil, democratic conversations on the web.
Many websites and social media platforms have ways of weeding it out through moderation and the ability to report such speech, and now Google has developed a tool to detect toxic speech.
Google and co-creator Jigsaw have partnered with media organisations including the Guardian and the New York Times to develop Perspective, which uses machine learning to gauge how toxic a comment is perceived to be.
Moderators could use this rating as part of their job or it could be used to empower readers to use a sliding scale to weed out a certain level of toxicity before they peruse the comments section.
Although Perspective is still in the development stage, it is possible to cut and paste in text and try it out now. Developers can also request access to the API.
