Sentiment Analysis of Tweets Including Emoji Data

Document Type

Conference Proceeding

Publication Date

12-4-2018

Abstract

Sentiment analysis of text is a valuable tool used to identify and classify bodies of text for various purposes, including public sentiment detection in political campaigns, spam detection and threat assessment. In this paper, we examine the effectiveness of incorporating Emoji data for Twitter data emotion classification. We conducted experiments using Multinomial Naïve Bayes (MNB) and Support Vector Machine (SVM) classification methods, with automatically labeled Twitter data. We compare the accuracy of these classification methods with and without the Emoji data included over varying vocabulary sizes. We find that MNB outperforms SVM on the data for large vocabulary sizes, and both classifiers perform slightly better with the Emoji data included.

Publication Source (Journal or Book title)

Proceedings - 2017 International Conference on Computational Science and Computational Intelligence, CSCI 2017

First Page

793

Last Page

798

This document is currently not available here.

Share

COinS