Electronic Theses and Dissertations





Document Type


Degree Name

Master of Science


Electrical and Computer Engr


Computer Engineering

Committee Chair

Mohammed Yeasin

Committee Member

Eddie Jacobs

Committee Member

Andrew Olney


We present a framework for mining the contents of smart phone application reviews to help both the designers and the users. Our question is: How do we incorporate users' views and opinions to improve the product design and users' experience. In particular we focus on the reviews of the applications that are designed for the people who are blind or visually impaired. The key idea is to use linguistic constructs to analyze the contents of the reviews using aspects, sentiments, and emotions. The framework has three components - aspect finder, sentiment analyzer, and opinion detector. Using unsupervised machine learning techniques, the aspect finder extracts application specific aspects from the reviews. The sentiment analyzer and opinion detector work together to determine whether a review is "Emotional" or "Rational". We evaluate the framework using crowd annotated data - though the performance of the aspect finder is not as expected, the F1-score of the review classifier is 0.7082 which is comparable to human evaluation. Such a system will complement the existing star rating systems by overcoming its limitations.


Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.