Masters Theses

Abstract

"Current sentiment analysis methods focus on determining the sentiment polarities (negative, neutral or positive) in users’ sentiments. However, in order to correctly classify users’ sentiments into their right polarities, the strengths of these sentiments must be considered. In addition to classifying users’ sentiments into their correct polarities, it is important to determine the sources and topics under which users’ sentiments fall. Sentiment strength helps as to understand the levels of customer satisfaction toward products and services. Sentiment topics on the other hand, helps to determine the specific product/service areas associated with user sentiments. This paper proposes two sentiment analysis approaches. First an approach which determines the sentiment strength expressed by consumers in terms of a scale (highly positive, +5 to highly negative, -5) is proposed. The approach includes a novel algorithm to compute the strength of sentiment polarity for each text by including the weights of the words used in the texts. Second, a sentiment mining approach which detects sentiment topic from text is proposed. The approach includes a sentiment topic recognition model that is based on Correlated Topics Models (CTM) with Variational Expectation-Maximization (VEM) algorithm. Finally, the effectiveness and efficiency of these models is validated using airline data from Twitter and customer review dataset from amazon.com"--Abstract, p. iii

Advisor(s)

Siau, Keng, 1964-

Committee Member(s)

Hilgers, Michael Gene
Nah, Fiona Fui-Hoon, 1966-

Department(s)

Business and Information Technology

Degree Name

M.S. in Information Science and Technology

Publisher

Missouri University of Science and Technology

Publication Date

Fall 2023

Pagination

viii, 49 pages

Note about bibliography

Includes bibliographical references (pages 43-48)

Rights

© 2015 Esi A. R. Adeborna, All Rights Reserved

Document Type

Thesis - Open Access

File Type

text

Language

English

Thesis Number

T 12313

Electronic OCLC #

1459757445

Download

Included in

Business Commons

Share

 
COinS