Personality Traits Prediction using DISC

This study describes a novel strategy for optimizing the cultural fit assessment phase in firm hiring procedures that combines natural language processing (NLP) techniques and semi-supervised self-training learning models. Using an unlabeled dataset taken from tweets, we used preprocessing approaches and word2vec embeddings to capture semantic commonalities in textual data. K-means clustering was then used to determine the best cluster count (k=4). Working with psychology experts, we classified 10% of the dataset with the DISC psychology model to improve model training. Using semi-supervised learning, we created a hybrid model that combined Convolutional Neural Networks (CNN) and Long Short-Term Memory networks (LSTM) to efficiently use both labeled and unlabeled data. Our research seeks to streamline the cultural fit assessment in hiring procedures, allowing for more rapid and accurate candidate evaluations.