On Thursday, Google’s Jigsaw unit released its artificial intelligence based troll detecting software tool named Perspective to battle online trolling. Perspective scans content and comments posted online and accordingly rates how toxic they are based on the ratings given by the public.
Google’s parent company, Alphabet has used machine learning technology to weed out toxic abuse of online comments.
According to Jared Cohen, president of Jigsaw, a technology incubator within Alphabet, this new tool allows people to decide whether they need to participate in a conversation or not.
The tool can be used by publishers of news sites in order to monitor their comments box.
Cohen added that the troll detecting tool is in the work-in-progress mode and they wanted people to participate so that it can be improved further. Now that the Perspective application program interface is available for developers and the public, it can be used to detect abuse and harassment online.
Teaching Perspective the difference between good and evil
Jared Cohen added that, he and his team reviewed hundreds of thousands of comments to figure out those comments/posts that are deemed toxic or abusive in a conversion. Based on this research, Perspective provided a score from 0 to 100 to new comments according to its similarity with comments classified as toxic.
Cohen’s team worked with Wikipedia, The New York Times and other major information outlets to develop this troll detecting software. The New York Times has 14 moderators who work round the clock to manually review every comment posted on their site in order to weed out inappropriate comments. Based on the comments archive provided by The Times, Jigsaw prepared an algorithm for running the tool Perspective.
Learning process of the troll detecting tool continues
There are some nuances that the tool is yet to pick up. Cohen further added that the tool, during its initial run, may mark some false positives, but during the course of time it will improve the efficiency once they get access to more comments.
Twitter also launched a new troll tool recently
Twitter has always been considered a playing ground for abusers and trolls. Recently, they launched a new tool to cut trolling. The company has said that they are planning to use existing data to help prevent users who have harassed others on Twitter from opening a new account after deleting an existing one. Twitter also said that they are making some changes to their algorithm to flag tweets that are considered abusive or are low in quality.
Hack Harassment, the group founded by Vox Media, Intel and Born this Way Foundation is trying to create awareness about facing harassment online.
Cohen concluded his speech by adding that Google’s troll detecting software’s mission is to handle the geopolitical challenges such as cyber security and online censorship.
