Prediction of personality traits from text using time efficient preprocessing and deep convolution neural network

Abstract

Persons express their sentiments as part of day-by-day communiqué. Personality traits reveals persons thoughts, about their feelings and persons behavior. Hence personality traits sets psychology how one person different from one another. Person thoughts expressed by what he/she write about any situation. Henceforth personality traits prediction is the vital research area. This research area belongs to NLP (Natural Language Processing), the most widely accepted of these traits are as: Openness (OPE), Conscientiousness (CON), Extraversion (EXT), Agreeableness (AGR), and Neuroticism (NEU). In this paper we have proposed to use time efficient sentence tokenization algorithm, efficient text preprocessing prominence on emoji’s followed by CNN deep learning classifier, proposed prediction model uses convolution filter for feature selection, further we have compared prediction model with machine learning based prediction model. We have also compared brute force tokenization method with proposed tokenization algorithm over dataset different in size.