Fuzzy-based machine learning for predicting narcissistic traits among Twitter users.

Abstract

Social media has provided a platform for people to share views and opinions they identify with or which are significant to them. Similarly, social media enables individuals to express themselves authentically and divulge their personal experiences in a variety of ways. This behaviour, in turn, reflects the user’s personality. Social media has in recent times been used to perpetuate various forms of crimes, and a narcissistic personality trait has been linked to violent criminal activities. This negative side effect of social media calls for multiple ways to respond and prevent damage instigated. Eysenck's theory on personality and crime postulated that various forms of crime are caused by a mixture of environmental and neurological causes. This theory suggests certain people are more likely to commit a crime, and personality is the principal factor in criminal behaviour. Twitter is a widely used social media platform for sharing news, opinions, feelings, and emotions by users. Given that narcissists have an inflated self-view and engage in a variety of strategies aimed at bringing attention to themselves, features unique to Twitter are more appealing to narcissists than those on sites such as Facebook. This study adopted design science research methodology to develop a fuzzy-based machine learning predictive model to identify traces of narcissism from Twitter using data obtained from the activities of a user. Performance evaluation of various classifiers was conducted and an optimal classifier with 95% accuracy was obtained. The research found that the size of the dataset and input variables have an influence on classifier accuracy. In addition, the research developed an updated process model and recommended a research model for narcissism classification.