Extracting Personality Traits From Digital Identity on Social Media

Abstract

This research investigates the potential of extracting personality traits from social media linguistic productions, focusing on Moroccan users' digital identity. Using an experimental exploratory methodology, the study applies IBM Watson's Personality Insights—grounded in the Big Five model—to fourteen Facebook and Twitter profiles. Data was gathered via netnography between 2020 and 2021 from accounts selected for rich expression and accessibility. Results confirm the feasibility of detecting traits like openness and conscientiousness, though accuracy varies; Arabic texts showed weaker correlations (mean absolute error – 0.09–0.12) than English. The study reveals that specific linguistic choices significantly influence algorithmic outputs, raising questions regarding potential manipulation. Ultimately, this novel application of AI to Moroccan bilingual data underscores the promise of digital psychology while emphasizing the critical need for enhanced algorithmic precision and contextualized interpretation in future research.