LLM predicts human behavior: A BERT-based approach for conscientiousness personality trait detection from online...

Area of Science:

• Artificial Intelligence
• Computational Linguistics
• Psychology

Background:

• Deep learning and Natural Language Processing (NLP) have advanced machine understanding of human language.
• User-generated content on digital platforms provides rich data for personality assessment.
• Predicting personality traits from social media has significant applications in psychology, healthcare, marketing, and education.

Purpose of the Study:

• To predict the conscientiousness personality trait using AI.
• To analyze individual perception as organized, dependable, and goal-oriented behavior.
• To evaluate the effectiveness of BERT against traditional Machine Learning (ML) and Deep Learning (DL) methods for personality detection.

Main Methods:

• Utilized the MBTI dataset for empirical analysis.
• Employed the BERT (Bidirectional Encoder Representations from Transformers) large language model.
• Compared BERT's performance with various ML and DL techniques incorporating feature engineering.

Main Results:

• BERT achieved the highest accuracy of 97% in predicting the conscientiousness trait.
• BERT demonstrated superior performance compared to other ML and DL approaches.
• Feature engineering techniques were used to extract textual and deep features for analysis.

Conclusions:

• BERT is a highly effective predictor for personality detection from textual data.
• This research offers valuable insights into understanding user behavior on social channels through personality analysis.
• The findings highlight the potential of advanced NLP models in psychological and behavioral assessments.