Using Text Mining to Compare Online Pro- and Anti-Vaccine Headlines: Word Usage, Sentiments, and Online Popularity

Abstract

This study aims to explore differences between health misinformation and true information by comparing word usage, sentiments, and online popularity between pro- and anti-vaccine headlines (PVHs and AVHs). Text mining and sentiment analysis showed that AVHs were more likely to use negative sentiment words and trust-related words. PVHs were more likely to use words related to positive sentiments. Anti-vaccine messages (AVMs) were more popular online than pro-vaccine messages (PVMs). AVMs’ online popularity was not related to its emotion words usage. Among PVMs, those with more positive sentiment words were more likely to be shared, commented on, and reacted to online. Wordclouds and word networks were created to visualize the word usage and clustering. Future directions regarding message design and automatic detection and analysis techniques are provided.

Keywords:

• Anti-Vaccine
• Misinformation
• Sentiment Analysis
• Text Mining
• Vaccine

Notes

_1. ShareCount (https://www.sharedcount.com) is an online service which shows the number of times an URL has been shared online according to its website. Twitter data was not supported by this service.