Personality Assessment Based on Natural Stream of Thoughts Empowered with Machine Learning

Knowing each other is obligatory in a multi-agent collaborative environment. Collaborators may develop the desired know-how of each other in various aspects such as habits, job roles, status, and behaviors. Among different distinguishing characteristics related to a person, personality traits are an effective predictive tool for an individual’s behavioral pattern. It has been observed that when people are asked to share their details through questionnaires, they intentionally or unintentionally become biased. They knowingly or unknowingly provide enough information in much-unbiased comportment in open writing about themselves. Such writings can effectively assess an individual’s personality traits that may yield enormous possibilities for applications such as forensic departments, job interviews, mental health diagnoses, etc. Stream of consciousness, collected by James Pennbaker and Laura King, is one such way of writing, referring to a narrative technique where the emotions and thoughts of the writer are presented in a way that brings the reader to the fluid through the mental states of the narrator. Moreover, computationally, various attempts have been made in an individual’s personality traits assessment through deep learning algorithms; however, the effectiveness and reliability of results vary with varying word embedding techniques. This article proposes an empirical approach to assessing personality by applying convolutional networks to text documents. Bidirectional Encoder Representations from Transformers (BERT) word embedding technique is used for word vector generation to enhance the contextual meanings.