Predicting Personality Traits from Instagram Captions Using NLP by Dauren Omarbekov

Personality plays a crucial role in shaping human behavior, preferences, and communication styles. With the rise of social media, individuals often express their thoughts, emotions, and personality traits through captions and posts. This project aims to predict personality traits - specifically those described by the Big Five Model called OCEAN (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism) - based on Instagram captions using Natural Language Processing techniques. Traditional personality assessments rely on self-reported questionnaires, which can be biased and impractical on scale. In contrast, analyzing social media content provides a passive and scalable alternative for personality inference. For this research, a dataset of Instagram captions labeled with personality scores was used. The project compares the performance of classical machine learning models such as Logistic Regression and Random Forest with deep learning models, particularly BERT (Bidirectional Encoder Representations from Transformers), for text classification. The preprocessing steps included tokenization, stopword removal, and embedding techniques for model input. Evaluation metrics such as accuracy, F1-score, and confusion matrices were used to assess the performance of each model. Results indicate that transformer-based models outperform traditional algorithms, achieving higher accuracy and better generalization across personality dimensions. This study demonstrates that NLP techniques can effectively be applied to social media text to infer psychological characteristics, offering potential applications in personalized marketing, mental health screening, and human-computer interaction. Future work may involve multimodal analysis incorporating images and extending the approach to other platforms like Twitter (X) or TikTok.