Comparative Analysis of Machine Learning based Classification Algorithms for Sentiment Analysis

Abstract

Sentiment analysis is the process of predicting the sentiment polarity of review data based on a given data set. Nowadays, sentiment analysis is more popular in Internet in general and in social media in particular. In the web huge amount of review data generated in each day is rapidly increasing day to day so there need to process these data to detect the sentiment polarity of large review dataset as early as possible. In this research, comparison of three different machine learning based classification algorithms for sentiment analysis i.e. Multinomial naïve Bayes (MNB), K-Nearest-Neighbors (KNN) and Support Vector Machines (SVM) is presented. The main aim of this research is to evaluate their performance of those three different machine learning based classification algorithms for sentiment labeled sentences datasets with different size. The sentiment labeled sentences datasets used for this research is chosen such way that they are different in size, mainly in terms of number of instances. When comparing the performance of all three machine learning based classification algorithms for sentiment analysis, SVM is found to be better algorithm to detect sentiment polarity in all three sentiment labeled sentence datasets in every aspect, whereas MNB and KNN had got less performance in every aspect as compared to SVM. Keywords: Sentiment analysis, KNN, SVM, MNB.