Date of Award

12-2016

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Computer Engineering and Sciences

First Advisor

Carlos Otero

Second Advisor

Ivica Kostanic

Third Advisor

Munevver Subasi

Fourth Advisor

Samuel Kozaitis

Abstract

In mobile system environments, the quality perceived by users is inconstant and reliant on many factors such as cellular network, data connection, cost, coverage area, etc. Even though Quality of Service (QoS) management is enabled in most modern telecommunication systems, it does not guarantee the actual user’s perceived Quality of Experience (QoE) level. Many cellular networks rely on engineering test research, such as drive testing or smart mobile applications, to collect the required parameters in order to provide better service quality to users. However, this approach does not always yield customer satisfaction. Hence, user opinions should be considered. These opinions can be found via social media, and collected and processed via social media analytics models. In this thesis research, a Rule-based algorithm is implemented. Based on this Rulebased algorithm, a sentiment analyzer is designed and tested. The results from testing the Rule-based algorithm are compared with results from a Naïve Bayes analyzer. In this thesis, the carrier Verizon is considered as the main topic and Twitter is considered the data source. This Rule-based algorithm and analyzer introduces a new method for generating datasets to easily design sentiment models. These models will analyze users’ opinions to make better decisions and recommend the optimal QoE solutions. The results of research conducted in this thesis show that the Rule-based analyzer performs better than the Naïve Bayes analyzer.

Download

Share

COinS