Reproduce “Myers-Briggs personality classification from social media text using pre-trained language models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Vitor Garcia dos Santos, Ivandré Paraboni
Researchers from Universidade de São Paulo evaluated the effectiveness of fine-tuned BERT models for Myers-Briggs Type Indicator (MBTI) personality classification from social media text across seven languages. The study found that BERT consistently outperformed LSTM and logistic regression baselines, achieving F1 scores over 0.90 for several personality dimensions and demonstrating strong cross-lingual generalization.
Key insights
- Pre-trained contextual language models like BERT significantly improve MBTI personality classification accuracy over traditional methods.
- Personality traits manifest differently in written language, leading to varying classification accuracies across MBTI dimensions (e.g., Extraversion/Introversion and Judging/Perceiving often showed higher accuracy).
- The approach demonstrates strong cross-lingual generalizability, with BERT performing robustly across multiple non-English languages.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.