Personality analysis of the USA public using Twitter profile pictures

摘要

The personality of people may be possibly analyzed using the content they post using social media sites and their profile picture choice. In this paper, we analyze Twitter profiles images to predict personality of major categories of the USA public. For the analysis, we have distributed the USA people into five major categories (Political personalities, Sports Stars, Business Bodies, Hollywood figures, and General public). We have collected the Twitter profile data, last month tweets and stored them in MySQL data base. This collection of data sets is further analyzed with their profile images using features of color, image composition, image type, and their online behavior. We are predicting about their personality traits using their posts and online engagement on Twitter. For the interpretability, our analysis targeted aesthetic and facial features to explore personality traits. Our results predict the behavior of Twitter users based on the profile picture they choose. For instance, conscientious and agreeable users display positive emotions, while users having openness select aesthetic photos. Our experimental results and calculation proved that we have achieved 53.457 percent accurate prediction and encounter that personality prediction is harder for robust accuracy.