Paper Title
Extraversion Detection in Profile Pictures Using IBM Watson

Abstract
The start of the era of cognitive computing can be dated back to 2011. Since then, more and more cognitive service providers appeared on the market, making it easier to use the supercomputers for developers and researchers. [1] It takes only a couple of clicks to start parsing large amount of unstructured data or training deep neural networks, that can used later for visual recognition. Watson is the name of the the cognitive computing system developed by IBM, that has several subsystems including the Personality Insights, that can extract personality traits from text and the Visual Recognition, that can be taught to analyze pictures and recognize them. In this exploratory research, public posts from Twitter were sent to IBM Watson’s Personality Insights to retrieve the writers’ Big5 personality traits, then their profile pictures were filtered and sorted according to the owner’s extraversion personality trait and used as a training data for Watson’s Visual Recognition. This exploratory research proves, that a sample of 2000 users is enough to reach a reasonably high precision for highly extroverted people, in case of valid prediction. The trained neural network does not outperform human beings, but more accurate than other recent researches. Index terms - IBM Watson, personality trait detection, visual recognition, social network, neural network