University of Wisconsin researchers found a similar trend in their study of Twitter conversations after 59 mass shootings from 2012 to 2014. That research, which has not yet been published, analyzed 1.3 million tweets and 700 related hashtags, using machine learning to sort them into various categories, said political science professor Jon C. W. Pevehouse, who co-authored the study with Dhavan V. Shah, a journalism professor, and several others.