Machine learning approaches to sentiment analysis in online social networks

Abstract

The online social network presents the quantitative measure of the psychological behaviour of individuals and helps to analyse the generic standpoint of social or political issues. As the field of research in text mining, it follows a computational approach to determine the opinions, sentiments, and subjectivity of text and other expressions. Moreover, the majority of approaches try to model the syntactic information of words without considering sentiment. The present study gives a brief narration of different machine learning (ML) models used for sentiment analysis and also proposes an efficient modular approach to give precise accuracy in validating and testing the Twitter data. The objective is to solve the problems through evaluation and comparison of different methods based on accuracy and training time. The proposed model achieves an accuracy of 88.37% with minimum possible training time. Simulation study states an effective way in which dataset may be thoroughly analysed and implemented with a focus on further validation of sentiment dataset to make tweet sentiment analysis more accurate.