Personality Trait Classification from Online Text Using Hybrid Deep Learning Techniques

In modern works, people accidentally divulge a plethora of information, enabling personality characteristics to be extracted. Recently, there has been an emphasis on cognitive-based sentiment analysis using online social media text. While deep learning (DL) algorithms have shown excellent performance in text-based personality assessment, their effectiveness might be altered by the activation functions used such as Tanh and leaky RELU activation function outperformed sigmoid across all datasets. Conventional procedures have fallen short of producing desirable results. This study presents a novel hybrid deep learning model, a fusion of convolutional neural networks (CNN) and long short-term memory (LSTM). It demonstrates its effectiveness in identifying eight key personality traits (Introversion–Extroversion, Intuition–Sensing, Thinking–Feeling, Judging–Perceiving) in real-world Twitter textual data. Our algorithm, trained on the MBTI dataset, outperforms cutting-edge approaches in identifying user personality characteristics. The approach's effectiveness is supported by extensive statistical research. The findings help firms to make more educated human recruiting choices and to apply research-based best practices for improving policies, services, and products. Notably, the CNN + LSTM hybrid model has an exceptional accuracy of 96.8%, outperforming other models such as Random Forest, KNN, XG Boost, SVM, Logistic Regression, and Gradient Descent.