Graph Data Mining: Algorithm, Security and Application

Preface
	References
Acknowledgment
Contents
1 Information Source Estimation with Multi-Channel Graph Neural Network
	1.1 Introduction
	1.2 Related Work
		1.2.1 Information Diffusion Modeling
		1.2.2 Information Source Detection
		1.2.3 Graph Neural Network
	1.3 Preliminaries
		1.3.1 Problem Definition
	1.4 Multi-Channel Graph Neural Network
		1.4.1 Feature Indices of Input
			1.4.1.1 Structural Features
			1.4.1.2 Prior Knowledge Features
		1.4.2 Graph Convolutional Networks
		1.4.3 Architecture of MCGNN
		1.4.4 Loss Function
	1.5 Experiment
		1.5.1 Datasets and Experimental Setup
		1.5.2 Baselines and Evaluation Metrics
		1.5.3 Results on the Synthetic Networks
		1.5.4 Results on the Real-World Networks
	1.6 Conclusion
	References
2 Link Prediction Based on Hyper-Substructure Network
	2.1 Introduction
	2.2 Existing Link Prediction Methods
		2.2.1 Heuristic Methods
		2.2.2 Embedding-Based Methods
		2.2.3 Deep Learning-Based Models
	2.3 Methodology
		2.3.1 Problem Formulation
		2.3.2 Neighborhood Normalization
		2.3.3 HSN Construction
		2.3.4 HELP
	2.4 Experiment
		2.4.1 Datasets
		2.4.2 Link Prediction Methods for Comparison
		2.4.3 Evaluation Metrics
		2.4.4 Experimental Settings
		2.4.5 Link Prediction Results
		2.4.6 Parameter Sensitivity
	2.5 Conclusion
	References
3 Broad Learning Based on Subgraph Networks for Graph Classification
	3.1 Introduction
	3.2 Related Work
		3.2.1 Subgraph Networks
		3.2.2 Network Representation
		3.2.3 Broad Learning System
	3.3 Subgraph Networks
		3.3.1 First-Order SGN
		3.3.2 Second-Order SGN
	3.4 Sampling Subgraph Networks
		3.4.1 Sampling Strategies
			3.4.1.1 Biased Walk (BW)
			3.4.1.2 Spanning Tree (ST)
			3.4.1.3 Forest Fire (FF)
		3.4.2 Construction of S2GN
	3.5 BLS Classifier Based on S2GN
		3.5.1 BLS Classifier
		3.5.2 Classification Framework
	3.6 Experiment
		3.6.1 Graph Classification
		3.6.2 Datasets
		3.6.3 Network Representation
		3.6.4 SGN for Graph Classification
		3.6.5 S2GN for Graph Classification
	3.7 Computational Complexity
	3.8 Conclusion
	References
4 Subgraph Augmentation with Application to Graph Mining
	4.1 Introduction
	4.2 Related Work
		4.2.1 Graph Classification
			4.2.1.1 Graph Kernel Methods
			4.2.1.2 Embedding Methods
			4.2.1.3 Deep Learning Methods
		4.2.2 Data Augmentation in Graph Learning
	4.3 The Model Evolution Framework for Graph Classification
		4.3.1 Problem Formulation
		4.3.2 Subgraph Augmentation
			4.3.2.1 Random Mapping
			4.3.2.2 Motif-Similarity Mapping
		4.3.3 Data Filtration
		4.3.4 Model Evolution Framework
	4.4 Application of Subgraph Augmentation
		4.4.1 Graph Classification
			4.4.1.1 Experimental Setting
		4.4.2 Link Prediction
			4.4.2.1 Subgraph Extraction
			4.4.2.2 Experimental Setting
		4.4.3 Node Classification
			4.4.3.1 Subgraph Extraction
			4.4.3.2 Experimental Setting
		4.4.4 Experimental Results
	4.5 Conclusion
	References
5 Adversarial Attacks on Graphs: How to Hide Your Structural Information
	5.1 Background
	5.2 Adversarial Attack
		5.2.1 Problem Definition
		5.2.2 Taxonomies of Attacks
	5.3 Attack Strategy
		5.3.1 Node Classification
			5.3.1.1 NETTACK
			5.3.1.2 Meta Attack
			5.3.1.3 Experiment of Results
		5.3.2 Link Prediction
			5.3.2.1 Heuristic Attack
			5.3.2.2 Gradient-Based Attack
			5.3.2.3 Experiment of Results
		5.3.3 Graph Classification
			5.3.3.1 Hierarchical Reinforcement Learning Attack
			5.3.3.2 Experimental Result
		5.3.4 Community Detection
			5.3.4.1 GA-Based Q-Attack
			5.3.4.2 Experiment Result
	5.4 Conclusion
	References
6 Adversarial Defenses on Graphs: Towards Increasing the Robustness of Algorithms
	6.1 Introduction
	6.2 Adversarial Training
		6.2.1 Graph Adversarial Training
		6.2.2 SAT
	6.3 Graph Purification
		6.3.1 GCN-Jaccard
		6.3.2 GCN-SVD
	6.4 Robustness Certification
		6.4.1 Certifying Robustness for Graph Structure Perturbations
		6.4.2 Certifying Robustness for Node Attributes Perturbations
		6.4.3 Certifiable Robustness in Community Detection
	6.5 Structure Based Defense
		6.5.1 Penalized Aggregation GNN
		6.5.2 Robust Graph Convolutional Network
	6.6 Adversarial Detection
		6.6.1 Adversarial Detection on Node Classification
		6.6.2 Adversarial Detection on Graph Classification
			6.6.2.1 SGN Based Adversarial Detection
			6.6.2.2 Joint Adversarial Detection
	6.7 Summary of Defenses
	6.8 Experiment and Analyze
		6.8.1 Adversarial Training
		6.8.2 Adversarial Detection
	6.9 Conclusion
	References
7 Understanding Ethereum Transactions via Network Approach
	7.1 Introduction
	7.2 Ethereum Transaction Dataset
	7.3 Graph Embedding Techniques
		7.3.1 Factorization Based Methods
		7.3.2 Random Walk Based Methods
		7.3.3 Deep Learning Based Methods
		7.3.4 Other Methods
	7.4 The Proposed Method
		7.4.1 Basic Definition
		7.4.2 Temporal Biased Walk
		7.4.3 Learning Temporal Graph Embeddings
	7.5 Experiment
		7.5.1 Node Classification
			7.5.1.1 Evaluation Metrics
			7.5.1.2 Experimental Results
		7.5.2 Link Prediction
			7.5.2.1 Evaluation Metrics
			7.5.2.2 Experimental Results
	7.6 Conclusion
	7.7 Appendix
		7.7.1 Similarity Indices
	References
8 Find Your Meal Pal: A Case Study on Yelp Network
	8.1 Introduction
	8.2 Data Description and Preprocessing
	8.3 Link Prediction Methods
		8.3.1 Similarity Indices Assembly
		8.3.2 Variational Graph Auto-Encoder
	8.4 Experiments
	8.5 Experiment Setup
		8.5.1 Friends Recommendation
		8.5.2 Co-foraging Prediction
	8.6 Conclusion
	References
9 Graph Convolutional Recurrent Neural Networks: A Deep Learning Framework for Traffic Prediction
	9.1 Background
	9.2 Related Work
		9.2.1 Graph Analysis
		9.2.2 Traffic State Prediction
	9.3 Model
		9.3.1 Graph Convolutional Network
		9.3.2 Long Short-Term Memory
		9.3.3 Graph Convolutional Recurrent Neural Network
	9.4 Experiment
		9.4.1 Dataset
		9.4.2 Baselines
		9.4.3 Evaluation
		9.4.4 Evaluation
		9.4.5 Results of Experiments and Analyses
	9.5 Conclusion
	References
10 Time Series Classification Based on Complex Network
	10.1 Introduction
	10.2 Related Work
		10.2.1 Time Series Classification
		10.2.2 Mapping Methods
		10.2.3 Graph Classification
	10.3 Methods
		10.3.1 Circular Limited Penetrable Visibility Graph
			10.3.1.1 Circle System Equation
			10.3.1.2 Graph Construction through CLPVG
			10.3.1.3 Subgraph Network
		10.3.2 Automatic Visibility Graph based on GNN
			10.3.2.1 The Overall Framework
			10.3.2.2 Feature Extraction
			10.3.2.3 Feature Matrix of Graph
			10.3.2.4 Classification of Graphs
		10.3.3 Comparison with LPVG
	10.4 Experiments
		10.4.1 Datasets
		10.4.2 The Experimental Settings
		10.4.3 The Experimental Results
	10.5 Conclusion
	References
11 Exploring the Controlled Experiment by Social Bots
	11.1 Introduction
	11.2 Definition of Social Bots
	11.3 Application and Influence of Social Bots
		11.3.1 Application
		11.3.2 Influence
	11.4 Development Technology of Social Bots
		11.4.1 Internet Access Technology
			11.4.1.1 PC Side
			11.4.1.2 Browser-based Access
			11.4.1.3 Mobile
		11.4.2 Artificial Intelligence Foundation
		11.4.3 Network Science Theory
	11.5 Social Bots Detection
		11.5.1 Graph-based Detection Method
		11.5.2 Feature-based Detection Method
		11.5.3 Crowdsourcing Detection Method
		11.5.4 Mixed Use of Multiple Ways
	11.6 Social Bots and Social Network Control Experiment
		11.6.1 Online Social Network Controlled Experiment
		11.6.2 Application of Social Bots in Controlled Experiment
		11.6.3 Problems in Controlled Experiments by Social Bots
			11.6.3.1 High Technical Threshold
			11.6.3.2 Legal and Moral Issues
	11.7 Conclusion
	References