DSpace Repository

Text Mining for Sentiment Analysis of Tweets in Twitter: Smartphone Product Review

Show simple item record

dc.contributor.advisor Esichaikul, Vatcharaporn
dc.contributor.author Rana, Anubhav
dc.contributor.other Guha, Sumanta
dc.contributor.other Anutariya, Chutiporn
dc.date.accessioned 2017-07-17T07:47:56Z
dc.date.available 2017-07-17T07:47:56Z
dc.date.issued 2017-05
dc.identifier.uri http://www.cs.ait.ac.th/xmlui/handle/123456789/872
dc.description 74 p. en_US
dc.description.abstract Millions of posts on social media are related to product reviews of a company’s products. To understand how these products are performing, sentiment analysis which is an application of text mining can be applied to extract the sentiment from each post. To know how well their product is doing companies cannot expect consumers to fill out lengthy feedback forms or answer calls asking for customer feedback as it is time consuming and does not benefit the consumer. However, on social media sites like Twitter, consumers willingly give their opinion on products in a 140 characters tweet. Companies can take advantage of this situation and perform sentiment analysis on such tweets to know whether consumers like or dislike their product. This research focuses on predicting the sentiment value for tweets taken from twitter.com for category of flagship smartphones. The flagship smartphones for which data is collected in this research are Apple iPhone7, OnePlus 3T and the Samsung Galaxy S7. Four classification techniques, namely Naïve Bayes, SVM (Support Vector Machines), decision trees and KNN (K-Nearest Neighbors) for sentiment analysis are compared in order to find out the best performing technique to carry out sentiment analysis for flagship smartphones. Results show that SVM classification technique is the best performer for sentiment analysis for flagship smartphones and Naïve Bayes is the second best classifier when compared to the other techniques. Through the user interface, the results of the best performing technique for the selected smartphone is displayed. This result includes a graph depicting the predicted number of good and bad tweets compared to the actual number, a graph displaying the results of the best performing classifier based on the performance measures criteria and a table showing the predicted sentiment values for each tweet. en_US
dc.description.sponsorship AIT Fellowship en_US
dc.publisher AIT en_US
dc.subject Sentiment Analysis en_US
dc.subject Tweets en_US
dc.subject Smartphone product review en_US
dc.title Text Mining for Sentiment Analysis of Tweets in Twitter: Smartphone Product Review en_US
dc.type Research report en_US


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account