Sentiment Analysis of Tweet Messages using Hybrid Approach Algorithm

Abstract

Communication is a vital component of everyday life. Through technology via social media, communication becomes more dynamic generating huge volume of data. Each data represents sentiments toward a public issue. Sentiment analysis algorithms able classify whether positive, negative or neutral. This paper introduces a hybrid algorithm combining two lexicon-based algorithms namely SentiWordNet and VADER algorithms. Three algorithms were tested using different data sources. It achieved an accuracy of 88.83% which 21.44% improvement from most commonly used algorithm SentiWordNet.