Supervised Machine Learning with BERT for Content Analysis
Chris J. Vargo · 2024
This chapter explores the application of supervised machine learning, specifically using BERT, for content analysis in social media, focusing on classifying tweets about vaccines and autism. It discusses the limitations of traditional keyword spotting and the advantages of machine-learning models that learn from labeled data to predict content categories. The chapter details the process of preparing data, training the model, and evaluating its performance using metrics like accuracy, precision, recall, and F1 score. It concludes by emphasizing the importance of model selection, evaluation, and ethical considerations in deploying machine learning for social science research.