Recent research has identified links between Psychopaths and the language they use (Hancock et al 2011), with media reports suggesting that such knowledge could be applied to social networks in order help Law Enforcement Agencies expose "Psychopath killers' traits". This is the first public study to research Psychopathy in the context of social media.
This study explored the extent to which it is possible to determine Psychopathy, and other personality traits based on Twitter usage. This was performed by comparing self-assessment 'Dark Triad' (Psychopathy, Machiavellianism, Narcissism) and 'Big Five' (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism) personality traits with the Twitter information, usage and language of 2927 participants.
Results show that there are a number of statistically significant correlations between an individual's darker personality traits and their Twitter activity. We also identified links between users' attitudes to privacy, their personality traits and their twitter use. We will present the improvement gains possible through the use of machine learning for personality prediction and share the models and techniques employed.
In addition to presenting our results, this talk will provide an introduction into identifying psychopathic traits using the Hare Psychopathy Checklist (PCL-R), present the technical approaches to collecting, storing and analyzing Twitter data using Open Source technologies and discuss the current ethical, privacy and human rights concerns surrounding social media analysis, vetting and labeling.
We will conclude with two proof of concept works, the first using the visualization tool Maltego to explore how visual analysis could be used to identify potential troublemakers at events such a far right demonstrations; the second to look at how personality traits influence response and interaction with a benign Twitter Bot.
The results highlight that in certain contexts, personality prediction through social media can perform with a reasonably high degree of accuracy.