A two-stage deep neural model with capsule network for personality identification.

This article focuses on developing a two-stage deep neural network model for automatic personality identification based on textual data, specifically using the Myers–Briggs Type Indicator (MBTI) model. The proposed architecture integrates three types of features: word-level semantic representations via static embeddings (fastText) processed by a capsule neural network, document-level contextualized embeddings (BERT), and psychological statistical features derived from text. Experimental results on a Kaggle dataset of social media posts demonstrate that combining these features improves classification accuracy over state-of-the-art models, supporting the hypotheses that hybrid text representations and inclusion of psychological features enhance personality detection. The study also discusses limitations related to language-specific psychological features and suggests directions for extending personality identification to low-resource languages.