Analysis and prediction of personality traits using a self-generated database of Moroccan instagram users: impact of gender on image content and quantity on prediction accuracy

The sharing of social media data has provided a significant amount of information about users. This has led to a growing interest in predicting user personality from social network data. Predicting user personality has potential applications in advertising and content recommendation systems, and can be a valuable source of insights for professionals seeking to understand and target their audiences on social media. This study focuses on the Instagram application and aims to extract and analyze the personality of Instagram users based on the content of their photos. Three categories of features are extracted from images: visual, emotional, and content. To describe users’ personalities, we used the Big Five model. One of the main contributions of this study is the creation of a comprehensive database that includes 316 Moroccan Instagram users. This database is the first of its kind to belong to Moroccan users. Additionally, The study also analyzed the influence of personality traits and gender on photo content by independently assessing the personalities of female and male Instagram users. Moreover, the personalities of each gender were extracted separately, and the influence of the number of images on prediction accuracy was studied. The study evaluates prediction accuracy using the root mean square error (RMSE) on a [1,5] score scale. Commendable results were achieved across all personality traits, with standout performance in predicting conscientiousness for females (RMSE \(=\) $=$ 0.59) and openness for males (RMSE \(=\) $=$ 0.59) compared to previous studies.