Title: A lexicon weighted sentiment analysis approach on Twitter

Authors: Mohammad Javad Shayegan; Mehrdad Molanorouzi

Addresses: Department of Computer Engineering, University of Science and Culture, Tehran, Iran ' Department of Computer Engineering, University of Science and Culture, Tehran, Iran

Abstract: Sentiment analysis in social media has grabbed more considerable attention because the results of such studies are highly applicable in social, economic, and political contexts. This study aimed to present an approach for collecting data from Twitter while storing and analysing the data using the hadoop as a big data platform as well as a hybrid trial and error model using the Bayes theorem plus a dictionary of words for the sentiment analysis. This method classifies tweets in two positive and negative classes based on the probability of positive words and negative words. According to the results, the accuracy of the proposed approach boosted from 67% to 71%. Then a new idea was employed in form of a weighted dictionary to achieve a higher accuracy. As such. the accuracy of the proposed approach reached a rate of 78% according to the results of another analysis conducted on the same data.

Keywords: twitter sentiment analysis; TSA; big data; lexicon sentiment analysis; hadoop.

DOI: 10.1504/IJWBC.2021.116641

International Journal of Web Based Communities, 2021 Vol.17 No.3, pp.149 - 162

Received: 19 Dec 2019
Accepted: 06 Jan 2021

Published online: 09 Jul 2021 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article