Researchers at the California Institute of Technology (Caltech) have demonstrated that machine-learning algorithms can monitor online social media conversations as they evolve, which could one day lead to an effective and automated way to spot online trolling.
The Tech Xplore website cited Researcher Michael Alvarez as saying: "This is one of the things I love about Caltech: the ability to bridge boundaries, developing synergies between social science and computer science."
Prevention of online harassment requires rapid detection of offensive social media posts, which in turn requires monitoring online interactions. Current methods to obtain such social media data are either fully automated or rely on a static set of harassment-related keywords. But, neither method is very effective, said the researchers.
The research team has developed a new technique dubbed GloVe to discover keywords relevant to harassment.
The tool represents words in a vector space, where the distance between two words is a measure of their linguistic or semantic significance.
This model can be adopted to find other words and expressions used by harassers on social media.
GloVe also shows the extent to which certain keywords are related, like the link that can be detected between the words "woman" and "sexual".
This project aims to provide a powerful tool for social media, to combat harassment and bullying online.
