Analyzing Public’s Reaction towards Black Lives Matter Campaign using Machine learning-based Approach through Spark

Abstract

Since the end of May 2020, there is a massive wave of protestant in the United States addressed to the government regarding the case of police violence towards black people. On May 25, 2020, George Floyd, a 46-years-old black American man was killed during an arrest for allegedly using a counterfeit bill in Minneapolis. This study analyzes the public’s reactions towards the Black Lives Matter campaign using a supervised machine learning-based approach. The proposed model uses logistic regression with word vectors as its feature. The model classifies the public’s reaction represented by tweets and crawls using Spark Streaming into sentiment class, i.e. positive, negative and neutral. In addition, named entity recognition analysis was also conducted in this study. The aim is to find who else besides George Floyd whose rights have been fought for by the public. SparkNLP is used to build the logistic regressions model, sentiment analysis and named entity recognition. This study finds that most of the public tweets had a negative tone addressed to the Floyd incident specifically and to the violence towards black people in general. Another finding is that the campaign not only fought for George Floyd, but also fought for the other victims like Rayshard Brooks, Dominique Fells and Eric Garner.