A Systematic Literature Review of Machine Learning Methods for Predicting Election Results Based on Social Media Platforms

Joanna WOŹNIAK and Adrian P. WOŹNIAK

Military University of Technology, Warsaw, Poland

Abstract

To develop more effective methods of predicting election results, many scientists have turned to Machine Learning methods based on data collected from social media. We performed a systematic literature review on that subject to answer our research questions. The questions were focused around data sources (social media), features used, preprocessing methods and Machine Learning models used to perform election results prediction. We answered all questions and found that the most popular social media was Twitter. Most common were Sentiment Analysis methods based on tweet texts. Most popular models used were those based on Bayes’ theorem, Decision Trees and Support Vector Machines.

Keywords: Machine Learning, Systematic Review, Social Media.
Shares