Multi-agent system for privacy protection through user emotions in social networks

Abstract

In this research privacy and decision making in social networks are addressed through a multi-agent system, using a model of the temperament of users, taking into account their emotions through the messages they put on the social media and the visual information obtained from them. We use opinion mining from a social network and images from users to get the data and calculate a model of temperament based on the PAD model generated from images and in the polarity of the messages they write. For this reason we propose a method to calculate a temperament state based on the history of PAD values of the user and the history of text polarities. We use also a method that analyzes the sentiment expressed in a message and helps the user to make the decision of posting it or not.