Personality Traits Assessment: A Case of Study Using Text Mining Techniques

This paper presents a complete experience to solving a personality trait assessment problem using Text Mining techniques within the framework of Knowledge Discovery in Databases. The study involves collaboration between researchers from the fields of Computer Science and Psychology, highlighting the interdisciplinary nature of the work. In this study, four basic predictive algorithms were evaluated: Multinomial Naive Bayes, Logistic Regression, Support Vector Machines, and Decision Trees. These algorithms were applied to address the classification problem posed by personality trait assessment. Given the nature of the problem, where individuals may possess multiple personality traits to varying degrees, the classification task was modeled in three different ways: binary, multiclass, and multilabel. To enhance the performance of the classification approaches, a data augmentation technique was employed. The results indicate that data augmentation improves the performance of all classification approaches, with binary classification benefiting the most. Moreover, for three out of the five personality traits studied, the weighted- \(F_1\) scores exceed 0.75, indicating strong predictive accuracy. Notably, the Responsibility trait achieves the highest score of 0.88, demonstrating the effectiveness of the classification approach for this particular trait.