Society is becoming increasingly more interconnected through the internet as a pipeline for sharing feelings and reactions to the world at large. In no domain is this more prevalent than in communities’ reactions to violent events. These events not only shake a community to its core but explode out onto the internet as reactions occur across the globe. This paper proposes two goals. The first goal is the refinement of sentiment analysis on tweets related to mass shooting events. The Python module Empath is used as an open source alternative to tools such as the LIWC and multiple classification algorithms are compared for accuracy using known, labeled tweet sets. The model selected by the paper is then applied to tweets collected around the time of 16 high profile shootings, ranging from the Aurora Theater shooting in in 2012 to the Santa Fe High School shooting in 2018. The paper’s second goal is to present a mapping of changing sentiment over time in regarding these events. This shall act to identify a trend in changing reactions toward shootings. The paper finishes with a discussion of what future expansion the work could undergo, as well as what other domains mapping sentiment over time can be a novel and useful exercise.
Empath, Sentiment Analysis, Artificial Intelligence, Machine Learning, Social Network Mining
Technologies and Human Usability
Paper Presentation in a Themed Session
Continuing Lecturer, Computer Science, Purdue Fort Wayne, United States
Gabriel Quinones Sanchez
Graduate Teaching Assistant, Purdue Fort Wayne, United States
Student, Purdue Fort Wayne, United States