Cyberbullying on social networking sites can cause harm by increasing the likelihood of suicidal thoughts and attempts and should be controlled. To control cyberbullying in the long term, it is necessary to change the attitudes and behaviors of individuals who engage in harmful communication. In the current study, we focused on positive affect (PA) and negative affect (NA), which are related to attitudes and behaviors. We developed two SNS applications (X-type and VRChat-type) that incorporate nudge using Multilingual Translation (NMT), which translates cyberbullying words into positive words in other languages. The process also exposes users to positive reactions from imaginary third parties who have seen these positive words. In this paper, we describe the development of two types of SNS applications and report the initial results of a pilot study to explore their potential utility.

Slander on social networking sites has become a social problem, and measures, such as legislation and stricter regulations have been taken. However, there are limits to these measures, and we have not seen any system that encourages fundamental changes of the mind and behavior of contributors. Therefore, we propose a training robot that contributes to slander suppression, targeting people who post negative words regularly. To change his/her mind and behavior, the proposed robot system will serve as an intermediary to help people break out of the addiction of feeling good about slander by converting negative words in negative messages into positive words, generating positive messages, and presenting them to the user. In this report, we will describe the concept and the initial development of the proposed system to be implemented in a real robot system.