Title Page
Abstract
Contents
Chapter 1. Introduction and Background 10
1.1. Introduction 10
1.2. Background 10
1.3. Problem Statement 11
1.4. Goals and Objectives 12
1.5. Structure of the Thesis 12
Chapter 2. Literature Review 13
2.1. Overview 13
2.2. Frequency-Based Techniques 13
2.3. Sentiment Analysis 14
2.3.1. Unsupervised Sentiment Analysis 14
2.3.2. Supervised Sentiment Analysis 15
2.4. Aspect- Based Sentiment Analysis 15
2.4.1. Benefits of Aspect- Based Sentiment Analysis 16
2.4.2. Four Sentiment Elements of ABSA 16
2.5. Datasets and Data Sources 17
2.6. Pre-processing Techniques 17
2.7. Feature Extraction 18
2.7.1. Forum Specific Features 19
2.7.2. Lexicon Features 19
2.8. Evaluation 20
Chapter 3. Research Methodology 21
3.1. Overview 21
3.2. Proposed Methodology 21
3.3. Acqusition of Data 22
3.4. Classification of Methodology 23
Chapter 4. Experimentation and Results 24
4.1. Overview 24
4.2. Dataset 24
4.3. Preprocessing 25
4.4. Analysis 25
4.5. Experimentation and Results 29
Chapter 5. Conclusions and Future Works 33
5.1. Conclusions 33
5.2. Future Works 34
References 35
Table 2.1. Forum Specific Entities and Features 19
Table 4.1. Cluster and Significant Terms 29
Table 4.2. Results Comparison 31
Figure 2.1. Sentiment Lexicons 20
Figure 3.1. Pipeline of the Proposed Methodology 22
Figure 3.2. Algorithm 23
Figure 4.1. LDA Topics and View Counts 26
Figure 4.2. NMF Topics and View Counts 27
Figure 4.3. Graphical Representation of Clusters 28
Figure 4.4. Dependency Parse Tree of a Post-Title Text 29
Figure 4.5. Demonstration of the system 30