Exploring how the usage of moral emotional words shapes the diffusion of political content on social networks using NLP techniques.
Authors: Hui Wen Goh and Yara Kyrychenko.
Ever since the 2016 US elections, we have been aware of how the spread of political ideas on social media can shape real-life outcomes. Yet, we still do not understand precisely how and why some political ideas spread more widely than others. In their 2017 paper, Brady et al. made one of the few groundbreaking discoveries on the topic. In this work, we replicate and extend their findings on moral-emotional contagion on social media. We studied the effects of positive and negative moral emotions on the diffusion of ideas using a large set of tweets on gun control. We further investigated the relationship between message spread and its complexity using the Gunning Fog Readability Index. We found that each additional moral emotional word increases retweet count by about 33%, while a unit increase in complexity produces only 1% decrease. We also found an asymmetric effect of positive versus negative moral-emotional words on message spread.
Check out our poster or watch our presentation at Westchester Undergraduate Research Conference 2021 to learn more.