Scholarlink

The Myers-Briggs Type Indicator (MBTI) classifies personality into 16 types across four dichotomies: Introversion-Extroversion (IE), Intuition-Sensing (NS), Feeling-Thinking (FT), and Judging-Perceiving (JP). This study predicts MBTI types from 8675 social media posts using machine learning (ML) and transformer models. After preprocessing text data, five ML models—Random Forest, XGBoost, SGD, Logistic Regression, and KNN—were trained, with XGBoost selected for its 58.40% accuracy and generalizability. Hyperparameter tuning improved XGBoost’s performance to 75.68% (IE) and 86.09% (NS). DistilBERT, constrained by system limitations, achieved 76.95% (IE) and 86.21% (NS). Enhanced with SMOTE, this approach targets over 80% accuracy for IE and NS, offering scalable personality profiling for applications in psychology and marketing. Validation on diverse texts (poem, cover letter, social media post) demonstrates practical utility.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Personality Prediction Using Social Media Interactions: A Machine Learning Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Harmandeep Singh,
  • Amdurkar Sanket Shivanand,
  • Divyashikha Sethia

摘要

The Myers-Briggs Type Indicator (MBTI) classifies personality into 16 types across four dichotomies: Introversion-Extroversion (IE), Intuition-Sensing (NS), Feeling-Thinking (FT), and Judging-Perceiving (JP). This study predicts MBTI types from 8675 social media posts using machine learning (ML) and transformer models. After preprocessing text data, five ML models—Random Forest, XGBoost, SGD, Logistic Regression, and KNN—were trained, with XGBoost selected for its 58.40% accuracy and generalizability. Hyperparameter tuning improved XGBoost’s performance to 75.68% (IE) and 86.09% (NS). DistilBERT, constrained by system limitations, achieved 76.95% (IE) and 86.21% (NS). Enhanced with SMOTE, this approach targets over 80% accuracy for IE and NS, offering scalable personality profiling for applications in psychology and marketing. Validation on diverse texts (poem, cover letter, social media post) demonstrates practical utility.