Detection of Content Incivility and its Causal Effects using Machine Learning
Date:
Using political incivility as an example, this talk presents a dictionary-based classification method, by leveraging word embeddings, could outperform supervised machine learning methods to detect uncivil content in online comments. Furthermore, this talk shows that machine learning models could help identify the causal effects of political incivility on engagement and contagions, by controlling for textual confounders.