Social media content has stirred much excitement and created abundant opportunities for understanding the opinions of the general public and consumers toward social events, political movements, company strategies, marketing campaigns, and product preferences. Many new and exciting social, geo political, and business-related research questions can be answered by analyzing the thousands, even millions, of comments and responses expressed in various blogs, forums, social media and social network sites, virtual worlds, and tweets. This is one of the good medium to explore the opinion of people about the particular event and so that this may help in the making any business decisions or the feedback about political activities to be carried out in future.
Therefore, we extracted the real time tweets on the social tweet keyword from the twitter web site, news website etc. using the twitter 4j Libraries and their API’s and JSOUP Libraries for obtaining the real time tweets from the respective web sites for English keyword only. These tweets are preprocessed and obtained the keyword related sentences only. These preprocessed tweets further used for the removal of
slang, hash, tags and URL and the removal of stop words. We also used the abbreviations and emoticon conversion to get corresponding complete meaning full message from tweets.
The processed tweets are further classified using three unstructured models EEC, IPC and SWNC. The results of these models are compared by obtaining the confusion matrix and their parameter such as precision, recall and accuracy. The SWNC model showed good result of classification over the EEC and IPC. Further the Hybrid model is used to reduce the number of the neutral tweets and obtained the corresponding results and shown by pie graph. By comparing the results of the SWNC model and the Hybrid model, it is observed that the numbers of neutral tweets are reduced in Hybrid model. The % range of reduction is around 20 - 25% in comparison with the SWNC model. Thus, we classified the real time social tweets using unstructured and their hybrid models. For obtaining these results, we developed the windows based indigenous, integrated and user friendly application in java and using NetBean’s framework.
Table of Contents
1. INTRODUCTION
1.1 Introduction
1.2 What is Social Network Analysis?
1.3 Motivation and Problem Statement
1.4 Objectives
1.5 Outline of Thesis
2. LITERATURE REVIEW
2.1 Introduction
2.2 Apriori-Based Approach
2.3 Pattern-Growth Approach
2.3.1 Survey and Techniques of Frequent Pattern Mining
2.4 Survey and Techniques of Online Social Networks
2.5 Survey on Opinion Mining
2.6 Summary
3. AN ANALYSIS OF SOCIAL NETWORK DATA USING GEPHI
3.1 Introduction
3.2 Features
3.2.1 Real-time Visualization
3.2.2 Layout
3.2.3 Metrics
3.2.4 Dynamic Network Analysis
3.2.5 Create Cartography
3.2.6 Clustering and Hierarchical graphs
3.2.7 Dynamic filtering
3.2.8 User-centric
3.2.9 Modular
3.2.10 Plug-in center
3.3 Implementation of Gephi to Social Network data
3.3.1 Data Collection and Experimental Details
3.4 Results and Discussion
3.5 Summary
4. TOOLS AND TECHNIQUES USED FOR EXTRACTION AND PROCESSING OF THE REAL TIME TWEETS
4.1 Introduction
4.2 Twitter 4j Library
4.2.1 System Requirements
4.2.2 How To Use
4.2.3 Download
4.2.4 Source Code
4.2.5 Maven Integration
4.3 Tweet search method by GET search/tweets in API V1.1
4.4 JSOUP JAVA Libraries
4.4.1 Parsing and traversing a Document
4.4.2 Parse a document from a String
4.4.3 Parsing a body fragment
4.4.4 Extract attributes, text, and HTML from elements
4.4.5 Sanitize untrusted HTML (to prevent XSS)
4.5 Netbeans IDE 8
4.5.1 Working with NetBeans Modules
4.6 Processing of tweets
4.7 SentiWordNet
4.7.1 History and team members
4.7.2 Database contents
4.7.3 Knowledge structure
4.7.4 Psycholinguistic aspects of WordNet
4.7.5 WordNet as a lexical ontology
4.7.6 Limitations
5. EXTRACTION AND THE PROCESSING OF REAL TIME TWEETS
5.1 Introduction
5.2 Experimental Details
5.3 Results and Discussions
5.4 Summary
6. CLASSIFICATION OF TWEETS USING SENTIMENTAL ANALYSIS
6.1 Introduction
6.2 Machine learning
6.2.1 Naive Bayes Classification
6.2.2 Maximum Entropy
6.3 Support Vector Machines
6.3.1 What SVM is used for?
6.3.2 How SVM Works
6.4 Evaluation of Sentiment Classification
6.5 Enhanced Emoticon Classification
6.5.1 EEC score calculation
6.6 Improved Polarity Classification
6.6.1 IPC score calculation
6.7 SentiwordNet Classification
6.7.1 SWNC score calculation
6.8 Method Details
6.9 Results And Discussions
6.10 Summary
7. CLASSIFICATION OF TWEETS USING HYBRID MODEL
7.1 Introduction
7.2 Methods
7.3 Results And Discussions
7.4 Summary
8. CONCLUSIONS AND FUTURE WORK
8.1 Conclusions
8.2 Future Work
9. REFERENCES
Research Objectives and Focus Areas
The research aims to analyze the evolution of social networks and extract meaningful sentiment from real-time Twitter data. The core research question addresses how social network analysis and sentiment classification algorithms, specifically unstructured models like Enhanced Emoticon Classifier (EEC), Improved Polarity Classifier (IPC), and SentiWordNet Classifier (SWNC), can be utilized in a hybrid framework to improve accuracy and reduce neutral sentiment classification in real-time tweet processing.
- Social network data analysis and visualization using the Gephi tool.
- Real-time data extraction from Twitter using Twitter4J and JSOUP libraries.
- Pre-processing techniques including slang removal, abbreviation expansion, and stop-word filtering.
- Development of a hybrid classification model to enhance sentiment detection accuracy.
- Comparative performance evaluation of sentiment classification models based on precision, recall, and accuracy metrics.
Excerpt from the Book
What is Social Network Analysis?
Social network analysis is based on assumption of the significance of relationships among interacting units. The social network perspective encompasses theories, models, and applications that are expressed in terms of relational concepts or processes. Along with growing interest and increased use of network analysis has become a consensus about the central principles underlying the network perspective. In addition to the use of relational concepts, we note the following as being important: Actors and their actions are viewed as interdependent rather than independent, autonomous units Relational ties (linkages) between actors are channels for transfer or "flow" of resources (either material or nonmaterial).
Network models focusing on individuals view the network structural environment as providing opportunities for constraints on individual action. Network models conceptualize structure (social, economic, political, and so forth) as lasting patterns of relations among actors. The unit of analysis in network analysis is not the individual, but an entity consisting of a collection of individuals and the linkages among them. Network methods focus on dyads (two actors and their ties), triads (three actors and their ties), or larger systems (subgroups of individuals, or entire networks.
Social network analysis has emerged as a set of methods for the analysis of social structures, methods which are specifically geared towards an investigation of the relational aspects of these structures. The use of these methods, therefore, depends on the availability of relational rather than attribute data.
Summary of Chapters
1. INTRODUCTION: This chapter introduces social network concepts, defines the research problem, and outlines the objectives of analyzing social network evolution and tweet sentiment.
2. LITERATURE REVIEW: This chapter provides an extensive survey of existing literature on graph mining, online social networks, and various sentiment analysis techniques.
3. AN ANALYSIS OF SOCIAL NETWORK DATA USING GEPHI: This chapter details the use of the Gephi tool for visualizing and analyzing structural properties like centrality, modularity, and degree distribution in social network datasets.
4. TOOLS AND TECHNIQUES USED FOR EXTRACTION AND PROCESSING OF THE REAL TIME TWEETS: This chapter covers the technical implementation using Twitter4J and JSOUP libraries for data extraction and preprocessing tasks.
5. EXTRACTION AND THE PROCESSING OF REAL TIME TWEETS: This chapter describes the practical experimental details and the specific methodology used to clean and process real-time Twitter data.
6. CLASSIFICATION OF TWEETS USING SENTIMENTAL ANALYSIS: This chapter explains the application of various classification models (EEC, IPC, SWNC) and evaluates their performance using standard statistical metrics.
7. CLASSIFICATION OF TWEETS USING HYBRID MODEL: This chapter proposes a hybrid approach to reduce neutral classifications and improve sentiment accuracy compared to standalone models.
8. CONCLUSIONS AND FUTURE WORK: This chapter summarizes the research findings and suggests potential future research directions, such as utilizing Hadoop frameworks for large-scale data analysis.
Keywords
Social Network Analysis, Sentiment Analysis, Twitter Mining, Gephi, Graph Theory, Machine Learning, Enhanced Emoticon Classifier, Improved Polarity Classifier, SentiWordNet, Data Preprocessing, Hybrid Models, Opinion Mining, Text Classification, Real-time Data Extraction, Social Media Analytics
Frequently Asked Questions
What is the primary focus of this research?
The research focuses on the study and analysis of social network evolution and the mining of public opinion using sentiment analysis on real-time Twitter data.
What are the core research themes?
The central themes include social network graph analysis, opinion mining in social media, development of sentiment classification models, and data preprocessing for real-time streams.
What is the primary objective of the work?
The primary objective is to analyze social networks through graph theory and to improve the accuracy of sentiment classification on tweets by developing a hybrid model that effectively reduces neutral sentiment results.
What scientific methods are utilized?
The study utilizes graph-based analysis with the Gephi tool, unstructured machine learning algorithms for sentiment classification (EEC, IPC, SWNC), and a hybrid framework to optimize classification results.
What does the main body of the work cover?
The main body covers the theoretical foundations of network analysis, technical procedures for real-time tweet extraction, implementation details for various sentiment classifiers, and a hybrid model for improved accuracy.
Which keywords define this work?
Key terms include Social Network Analysis, Opinion Mining, SentiWordNet, Hybrid Model, Twitter Mining, and Gephi.
Why is the hybrid model necessary?
The hybrid model is implemented because individual unstructured models frequently classify tweets as neutral; the hybrid approach significantly reduces this neutral output, leading to more definitive sentiment results.
What tools were developed for this project?
The author developed an indigenous, integrated, and user-friendly Windows-based application using Java and the NetBeans framework to automate the extraction, preprocessing, and classification of tweets.
- Quote paper
- Dr Bapurao Bandgar (Author), 2019, Social Networks and their Opinion Mining, Munich, GRIN Verlag, https://www.grin.com/document/512868
