Classification of Personality Traits on Facebook Using Key Phrase Extraction, Language Models and Machine Learning

The study of automated personality trait detection has become increasingly prevalent in the field of affective computing and sentiment analysis. Psycholinguistic databases and linguistic styles are found to have correlations with personality traits. Natural language processing has improved thanks to transfer learning. With pre-trained language models as features, this paper presents a machine learning model that benefits from transfer learning. The proposed method using key phrase extraction and SentenceTransformer through decision tree classifier surpasses the XGBoost by 26.79^% in terms of accuracy and exhibits 97.79% on myPersonality dataset.