Advances in Knowledge Discovery and Data Mining: 25th Pacific-Asia Conference, PAKDD 2021, Virtual Event, May 11–14, 2021, Proceedings, Part I (Lecture Notes in Computer Science, 12712)

General Chairs’ Preface
PC Chairs’ Preface
Organization
Contents – Part I
Contents – Part II
Contents – Part III
Applications of Knowledge Discovery
Fuzzy World: A Tool Training Agent from Concept Cognitive to Logic Inference
	1 Introduction
		1.1 Basic Idea
		1.2 The Proposed Tool
		1.3 Main Contributions
	2 Related Work
	3 Prelimilaries
		3.1 Deep Q-Learning Based on POMDP
		3.2 Concept Cognitive Task Formulization
		3.3 Logic Reasoning Task Formulization
	4 Fuzzy World and Baselines
		4.1 The Fuzzy World environment
		4.2 The Concept Cognitive Task and Baseline
		4.3 The Logic Reasoning Task and Baselines
	5 Experiments
		5.1 Experiment Tasks Based on Fuzzy World
		5.2 Experiment Details
		5.3 Experiment Results
	6 Conclusion
		6.1  Using Second Order Derivative Gradient for Cross Training Parameter
	References
Collaborative Reinforcement Learning Framework to Model Evolution of Cooperation in Sequential Social Dilemmas
	1 Introduction
	2 Mathematical Setup of CRL Model
		2.1 Extension of CRL to CDQN
	3 Related Work
	4 CRL Model on Repeated PGGs
		4.1 Experimental Results
		4.2 CDQN Model on Fruit Gathering Game (FGG)
		4.3 Experimental Results
	5 Conclusion
	References
SigTran: Signature Vectors for Detecting Illicit Activities in Blockchain Transaction Networks
	1 Introduction
	2 Background and Related Work
	3 Proposed Method
		3.1 Transactions History Retrieval
		3.2 Network Construction
		3.3 SigTran
		3.4 Node Classification
	4 Experiments
		4.1 Dataset Description
		4.2 Baseline Methods
		4.3 Performance Evaluation
	5 Conclusions
	References
VOA*: Fast Angle-Based Outlier Detection over High-Dimensional Data Streams
	1 Introduction
	2 Related Work
	3 Problem Definition
	4 Proposed Methods
		4.1 IncrementalVOA
		4.2 VOA*
	5 Experiments and Results
		5.1 Accuracy
		5.2 Performance
	6 Conclusion
	References
Learning Probabilistic Latent Structure for Outlier Detection from Multi-view Data
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Problem Definition
		3.2 Multi-view Bayesian Outlier Detector Formulation
		3.3 Multi-view Bayesian Outlier Detector Inference
		3.4 Multi-view Outlier Score Estimation
	4 Experimental Evaluations
		4.1 Experiment Setup
		4.2 Experiment Results
	5 Conclusions
	References
GLAD-PAW: Graph-Based Log Anomaly Detection by Position Aware Weighted Graph Attention Network
	1 Introduction
	2 Related Work
		2.1 Log-Based Anomaly Detection
		2.2 Graph Neural Networks
	3 The Proposed Model
		3.1 Problem Statement
		3.2 Graph Construction
		3.3 Position Aware Weighted Graph Attention Layer
		3.4 Session Embeddings
		3.5 Prediction and Anomaly Detection
	4 Experiments
		4.1 Datasets
		4.2 Baslines and Evaluation Metrics
		4.3 Experimental Setup
		4.4 RQ1: Comparison with Baseline Models
		4.5 RQ2: Comparison with Other GNN Layers
		4.6 RQ3: Parameter Sensitivity
	5 Conclusion
	References
CubeFlow: Money Laundering Detection with Coupled Tensors
	1 Introduction
	2 Related Work
	3 Problem Formulation
	4 Proposed Method
		4.1 Proposed Metric
		4.2 Proposed Algorithm: CubeFlow
	5 Experiments
		5.1 Experimental Setting
		5.2 Q1.Effectiveness
		5.3 Q2. Performance on Real-World Data
		5.4 Q3. Performance on 4-mode Tensor
		5.5 Q4. Scalability
	6 Conclusion
	References
Unsupervised Boosting-Based Autoencoder Ensembles for Outlier Detection
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Architecture and Training of Autoencoders
	4 Experiments
	5 Discussion
	6 Conclusion
	References
Unsupervised Domain Adaptation for 3D Medical Image with High Efficiency
	1 Introduction
	2 Related Work
		2.1 Domain Adaptation
		2.2 3D Medical Image Processing
	3 Methods
		3.1 Slice Subtract Module (SSM)
		3.2 LSTM Module
		3.3 Training of the Domain Adaptation
	4 Experiments
		4.1 Dataset and Experimental Settings
		4.2 Segmentation Evaluation
		4.3 Slices Difference Evaluation
		4.4 Cost Evaluation
	5 Conclusion
	References
A Hierarchical Structure-Aware Embedding Method for Predicting Phenotype-Gene Associations
	1 Introduction
	2 Method
		2.1 Hierarchical Structure-Aware Random Walk
		2.2 Node Embedding Learning with SkipGram
		2.3 Prediction of Phenotype-Gene Associations
	3 Experiments
		3.1 Data Preparation
		3.2 Baselines and Results
		3.3 Effectiveness of Hierarchical Structure-Aware Random Walk
		3.4 Analysis on Different Parameter Settings in SkipGram
		3.5 Predicting Causal Genes for Parkinson's Disease
	4 Conclusion
	References
Autonomous Vehicle Path Prediction Using Conditional Variational Autoencoder Networks
	1 Introduction
	2 Related Works
	3 Methods
	4 Experiment
	5 Conclusion
	References
Heterogeneous Graph Attention Network for Small and Medium-Sized Enterprises Bankruptcy Prediction
	1 Introduction
	2 Related Work
	3 Problem Formulation
	4 Methodology of Our Proposed Model
		4.1 Heterogeneous Neighborhood Encoding Layer
		4.2 Triple Attention and Output Layer
		4.3 Training
	5 Experiment
	6 Conclusion
	References
Algorithm Selection as Superset Learning: Constructing Algorithm Selectors from Imprecise Performance Data
	1 Introduction
	2 The Per-instance Algorithm Selection Problem
		2.1 Common Algorithm Selection Solutions
		2.2 The Influence of Censored Data
	3 Superset Learning with Right-Censored Data
	4 Experimental Evaluation
		4.1 Experimental Setup
		4.2 Baseline Approaches
		4.3 Results
	5 Conclusion
	References
Sim2Real for Metagenomes: Accelerating Animal Diagnostics with Adversarial Co-training
	1 Introduction
	2 Related Work
	3 Sim2Real Metagenome Sequence Classification
		3.1 Feature Representation Using De Bruijn Graphs
		3.2 The Problem with Using Pure Synthetic Data
		3.3 Adversarial Co-training: A Deep Learning Baseline
		3.4 Implementation Details
	4 Experimental Evaluation
		4.1 Data and Baselines
		4.2 Quantitative Evaluation
	5 Discussion and Future Work
	References
Attack Is the Best Defense: A Multi-Mode Poisoning PUF Against Machine Learning Attacks
	1 Introduction
	2 Related Work
	3 Multi-Mode Poisoning PUF
		3.1 Working Mode Design
		3.2 Hiding the Working Modes
		3.3 Deep Learning Attack
	4 Experimental Result
		4.1 Experimental Setup
		4.2 Performance of Working Modes Design
		4.3 Impact of Increasing the Number of Working Modes
		4.4 Impact of Imbalanced Working Modes
		4.5 Impact of Output Transition Probability
	5 Conclusion
	References
Combining Exogenous and Endogenous Signals with a Semi-supervised Co-attention Network for Early Detection of COVID-19 Fake Tweets
	1 Introduction
	2 Related Work
	3 Our Proposed Dataset: ECTF
	4 Our Proposed Methodology: ENDEMIC
		4.1 Exogenous Signals
		4.2 Endogenous Signals
		4.3 Connecting Components and Training
	5 Experimental Setup and Results
		5.1 Baseline Methods
		5.2 Evaluating on General-Test Set
		5.3 Evaluating on Early Detection
		5.4 Evaluating on General-Domain Fake News
	6 Conclusion
	References
TLife-LSTM: Forecasting Future COVID-19 Progression with Topological Signatures of Atmospheric Conditions
	1 Introduction
	2 Related Work
	3 Problem Statement
	4 Background
		4.1 Preliminaries on Topological Data Analysis
		4.2 Long Short Term Memory
	5 The Proposed Topological Lifespan Long Short Term Memory (TLife-LSTM) Approach
		5.1 Topological Features of Environmental Dynamic Networks
		5.2 Topological Machine Learning Methodology: TLife-LSTM
	6 Experiments
		6.1 Data, Experimental Setup and Comparative Performance
	7 Conclusions and Future Scope
	References
Lifelong Learning Based Disease Diagnosis on Clinical Notes
	1 Introduction
	2 Methodology
		2.1 Problem Definition and Nomenclature
		2.2 Lifelong Learning Diagnosis Benchmark: Jarvis-40
		2.3 Overall Framework
	3 Experiments
		3.1 Compared Baselines
		3.2 Dataset
		3.3 Experimental Protocol
		3.4 Overall Results on Benchmark Javis-40
		3.5 Importance of External Knowledge and Attention Mechanism
		3.6 Analysis of Entity Embeddings and Visualization
		3.7 Experiment Based on Medical Referral
	4 Conclusion
	References
GrabQC: Graph Based Query Contextualization for Automated ICD Coding
	1 Introduction
	2 Related Work
	3 GrabQC - Our Proposed Approach
		3.1 Named Entity Recognition and Entity Linking
		3.2 Query Extraction
		3.3 Generation of Contextual Graph
		3.4 Relevant Node Detection
		3.5 Distant Supervised Dataset Creation
	4 Experimental Setup
		4.1 Dataset Description
		4.2 Setup and Baselines
		4.3 Implementation Details
	5 Results and Discussion
		5.1 Performance of the Relevant Node Detection Model
		5.2 Query-Level Comparison
		5.3 Comparison with Machine Learning Models
		5.4 Analysis
	6 Conclusion
	References
Deep Gaussian Mixture Model on Multiple Interpretable Features of Fetal Heart Rate for Pregnancy Wellness
	1 Introduction
	2 Related Work
	3 Approach
		3.1 Data Preprocessing
		3.2 Extraction of Multiple Interpretable Features
		3.3 Interpretable Deep Gaussian Mixture Model
	4 Experiments
		4.1 Dataset
		4.2 Settings
		4.3 Results
		4.4 Interpretation of DGMM
		4.5 Discussion
	5 Conclusion
	References
Adverse Drug Events Detection, Extraction and Normalization from Online Comments of Chinese Patent Medicines
	1 Introduction
	2 Related Work
		2.1 ADE Mining in Social Media
		2.2 Multi-task Learning in ADE Discovery
	3 Dataset
	4 Model
		4.1 Problem Formulation
		4.2 Encoder and Primary Decoder
		4.3 Enhancing Decoder
		4.4 Output Layers
	5 Experiments
		5.1 Evaluation Metric
		5.2 Baselines and Scenarios
		5.3 Results for ADE Detection
		5.4 Results for ADE Extraction
		5.5 Results of ADE Normalization
	6 Conclusion and Future Work
	References
Adaptive Graph Co-Attention Networks for Traffic Forecasting
	1 Introduction
	2 Preliminaries
	3 Adaptive Graph Co-Attention Network
		3.1 Adaptive Graph Learning
		3.2 Hierarchical Spatio-Temporal Embedding
		3.3 Long- and Short-Term Co-Attention Network
	4 Experiment
		4.1 Datasets
		4.2 Baseline Methods
		4.3 Experiment Settings
		4.4 Experimental Results
		4.5 Conclusions
	References
Dual-Stage Bayesian Sequence to Sequence Embeddings for Energy Demand Forecasting
	1 Introduction
	2 Related Works
	3 Method
		3.1 Uncertainty Estimation
		3.2 Model Design
	4 Experimental Settings
		4.1 Description of the Datasets
		4.2 Feature Engineering and Selection
		4.3 Network Settings
	5 Experimental Results
		5.1 Prediction Performance
		5.2 Uncertainty Estimation
		5.3 Summary
		5.4 Detecting Anomalies in Energy Consumption
	6 Conclusion
	References
AA-LSTM: An Adversarial Autoencoder Joint Model for Prediction of Equipment Remaining Useful Life
	1 Introduction
	2 Proposed Approach
		2.1 Encoder Model
		2.2 Decoder Model
		2.3 Discriminator Model
		2.4 RUL Predictor
		2.5 Training Process
	3 Experiment
		3.1 Introduction to Turbofan Data Set
		3.2 Data Preprocessing
		3.3 Performance Metrics
		3.4 Results
	4 Conclusion
	References
Data Mining of Specialized Data
Analyzing Topic Transitions in Text-Based Social Cascades Using Dual-Network Hawkes Process
	1 Introduction
	2 Dual-Network Hawkes Process
	3 Experiments and Results
		3.1 Models Evaluated
		3.2 Datasets
		3.3 Evaluation Tasks and Results
		3.4 Results
		3.5 Analytical Insight from USPol Dataset
	4 Related Work
	References
HiPaR: Hierarchical Pattern-Aided Regression
	1 Introduction
	2 Preliminaries and Notation
	3 Related Work
	4 HiPaR
		4.1 Initialization
		4.2 Candidates Enumeration
		4.3 Rule Selection
		4.4 Prediction with HiPaR
	5 Evaluation
		5.1 Experimental Setup
		5.2 Comparison with the State of the Art
		5.3 Anecdotal Evaluation
	6 Conclusions and Outlook
	References
Improved Topology Extraction Using Discriminative Parameter Mining of Logs
	1 Introduction
	2 Related Work
	3 System Overview
		3.1 Template Mining
		3.2 Topology Extraction
		3.3 Online Module Usecase
	4 Experimental Results
		4.1 Invariant/Parameter Identification from Logs
		4.2 Template Extraction
		4.3 Application Topology Discovery
		4.4 Observations
	5 Conclusion and Future Work
	References
Back to Prior Knowledge: Joint Event Causality Extraction via Convolutional Semantic Infusion
	1 Introduction
	2 Related Work
		2.1 Pattern Matching
		2.2 Machine Learning
		2.3 Neural Network
	3 Our Model
		3.1 BERT Encoder
		3.2 Convolutional Semantic Infusion
		3.3 Query-Key Attention
		3.4 BiLSTM+CRF
		3.5 Training Objective and Loss Function
	4 Experiment
		4.1 Datasets
		4.2 Experimental Settings
		4.3 Results and Analysis
		4.4 Effectiveness of Semantic Infusion
	5 Conclusion
	References
A k-MCST Based Algorithm for Discovering Core-Periphery Structures in Graphs
	1 Introduction
	2 Related Work
	3 Our Approach
		3.1 Definitions
		3.2 Algorithm
		3.3 Complexity Analysis
	4 Experiments and Results on Real World Data Sets
		4.1 Primary School Data Set
		4.2 Airport Network
		4.3 Effect of Varying the Parameter k
		4.4 Validation of Use of k-MCSTs for CP Structures
	5 Conclusion
	References
Detecting Sequentially Novel Classes with Stable Generalization Ability
	1 Introduction
	2 Related Work
	3 Preliminaries
	4 Proposed Method
		4.1 The Framework of DesNasa
		4.2 Novel Class Detection
		4.3 Learn to Recognize
	5 Experiments
	6 Conclusion
	References
Learning-Based Dynamic Graph Stream Sketch
	1 Introduction
	2 Related Work
	3 Preliminaries
	4 Learning-Based Dynamic Graph Sketch Mechanism
		4.1 Learning-Based Error Detection
		4.2 CNN-Based Graph Sketch Expansion
	5 Performance Evaluation
		5.1 Experimental Environment
		5.2 Numerical Results
	6 Conclusion
	References
Discovering Dense Correlated Subgraphs in Dynamic Networks
	1 Introduction
	2 Problem Statement
	3 Solution
	4 Experimental Evaluation
	5 Related Work
	6 Conclusions
	References
Fake News Detection with Heterogenous Deep Graph Convolutional Network
	1 Introduction
	2 Methodology
		2.1 NDG Construction
		2.2 HDGCN
		2.3 Optimazation Objective
	3 Experiment
		3.1 Dataset
		3.2 Parameter Setting and Evaluation Metrics
		3.3 Baselines
		3.4 Performance Comparison
		3.5 Ablation Studies
		3.6 Further Analysis
	4 Related Work
	5 Conclusion
	References
Incrementally Finding the Vertices Absent from the Maximum Independent Sets
	1 Introduction
	2 Preliminary
	3 Baseline Method and Framework of Our Approach
		3.1 Baseline Brute-Force Method
		3.2 Framework of Our Approach
	4 Polynomial Algorithms
		4.1 Extended Domination Reduction
		4.2 Chain-Chased Reduction
	5 Experimental Results
	6 Conclusion
	References
Neighbours and Kinsmen: Hateful Users Detection with Graph Neural Network
	1 Introduction
	2 Related Work
	3 Problem Definition
	4 The HateGNN Framework
		4.1 Latent Graph Construction
		4.2 Biased Sampling Neighbourhood
		4.3 Multi-modality of Neighbourhood Aggregation
		4.4 Model Training
	5 Experiments
		5.1 Comparisons with Baselines (RQ1)
		5.2 Performance Analysis (RQ2)
		5.3 Parameter Analysis (RQ3)
	6 Conclusions
	References
Graph Neural Networks for Soft Semi-Supervised Learning on Hypergraphs
	1 Introduction
	2 Related Work
	3 Method
		3.1 Directed Hypergraph
		3.2 Soft SSL on Directed Hypergraphs
		3.3 Directed Hypergraph Network (DHN)
		3.4 The Supervised Loss L
	4 Theoretical Analysis: Generalisation Error
		4.1 Assumptions/Notations
		4.2 Definitions: Generalisation and Empirical Errros
		4.3 Extension to GNNs on Hypergraphs
	5 Experiments
		5.1 Experimental Setup
		5.2 Baselines
		5.3 Discussion
	6 Conclusion
	References
A Meta-path Based Graph Convolutional Network with Multi-scale Semantic Extractions for Heterogeneous Event Classification
	1 Introduction
	2 Related Work
		2.1 Meta-paths in HIN
		2.2 Graph Neural Networks
	3 Methodology
		3.1 Problem Definition
		3.2 Edge: Meta-path Based Similarity Measure
		3.3 Node Representation: Multi-scale Semantic Feature Extraction
		3.4 Local Extrema GCN Model
	4 Experiments
		4.1 Datasets and Settings
		4.2 Comparative Methods
		4.3 Results and Evaluation
		4.4 Sensitivity Analysis
		4.5 Interpretability
	5 Conclusions
	References
Noise-Enhanced Unsupervised Link Prediction
	1 Introduction
	2 Literature Review
	3 Noise Injection Methods
	4 Experimental Setup
	5 Theoretical Analysis
	6 Experimental Analysis
		6.1 Noise-Enhanced Link Prediction in Real-World Networks
		6.2 Noise-Enhanced Link Prediction in Synthetic Networks
	7 Conclusions
	References
Weak Supervision Network Embedding for Constrained Graph Learning
	1 Introduction
	2 Problem Definition
	3 CEL: Constraint Embedding Loss
	4 WSNE for Constrained Graph Learning
	5 Constraint Assisted Topology Optimization
	6 Experiments
		6.1 Constrained Graph Clustering
		6.2 Constrained Graph Classification
		6.3 Embedding Visualization
	7 Conclusion
	References
RAGA: Relation-Aware Graph Attention Networks for Global Entity Alignment
	1 Introduction
	2 Related Work
		2.1 TransE-Based Entity Alignment
		2.2 GCNs-Based Entity Alignment
		2.3 Global Entity Alignment
	3 Problem Definition
	4 RAGA Framework
		4.1 Basic Neighbor Aggregation Networks
		4.2 Relation-Aware Graph Attention Networks
		4.3 End-to-End Training
		4.4 Global Alignment Algorithm
	5 Experiments
		5.1 Experimental Settings
		5.2 Experimental Results and Analysis
	6 Conclusion
	References
Graph Attention Networks with Positional Embeddings
	1 Introduction
	2 Method
		2.1 The GAT-POS Model
		2.2 Positional Embedding Enhanced Graph Attentional Layer
		2.3 Positional Embedding Model
	3 Related Works
	4 Experiments and Results
		4.1 Datasets
		4.2 Methods in Comparison
		4.3 Experimental Setup
		4.4 Results
		4.5 Ablation Study
	5 Conclusion
	References
Unified Robust Training for Graph Neural Networks Against Label Noise
	1 Introduction
	2 Related Work
		2.1 Learning with Noisy Labels
		2.2 Graph Neural Networks
	3 Problem Definition
	4 The UnionNET Learning Framework
		4.1 Label Aggregation
		4.2 Sample Reweighting
		4.3 Label Correction
		4.4 Model Training
	5 Experiments
		5.1 Comparison with State-of-the-Art Methods
		5.2 Ablation Study
		5.3 Hyper-parameter Sensitivity
	6 Conclusion
	References
Graph InfoClust: Maximizing Coarse-Grain Mutual Information in Graphs
	1 Introduction
	2 Notation, Definitions, and Problem Statement
	3 Graph InfoClust (GIC)
		3.1 Motivation and Overview
		3.2 Coarse-Grain Loss
		3.3 Coarse-Grain Summaries
		3.4 Fake Input and Discriminators
	4 Related Work
	5 Experimental Methodology and Results
		5.1 Methodology and Configuration
		5.2 Results
	6 Conclusion
	References
A Deep Hybrid Pooling Architecture for Graph Classification with Hierarchical Attention
	1 Introduction
	2 Proposed Solution: HybridPool
		2.1 Overview of the Architecture
		2.2 GIN Embedding Layer
		2.3 GIN Pooling Layer
		2.4 Formulation of a Level Graph
		2.5 Attending the Important Hierarchies
		2.6 Run Time Complexity of HybridPool
		2.7 Variants of HybridPool: Model Ablation Study
	3 Experimental Evaluation
		3.1 Datasets and Baselines
		3.2 Performance Analysis for Graph Classification
	4 Discussion and Future Work
	References
Maximizing Explainability with SF-Lasso and Selective Inference for Video and Picture Ads
	1 Introduction
	2 Related Work
		2.1 Explainable Models
		2.2 Lasso and Its Variants
		2.3 Selective Inference for Lasso
	3 Intrinsic Explainable Linear Models
		3.1 Predictability Maximized Methods
	4 Proposed Explainability Maximized Method
		4.1 Selective Inference
	5 Empirical Evaluation
		5.1 Illustration of SF-Lasso
		5.2 Experiment Results
		5.3 Human Evaluations
	6 Conclusion
	References
Reliably Calibrated Isotonic Regression
	1 Introduction
	2 Background: Isotonic Regression
	3 Method
		3.1 Credible Intervals for Isotonic Regression
		3.2 Reliably Calibrated Isotonic Regression
		3.3 Greedy Optimization Algorithm
		3.4 Parameter Choice
	4 Related Work
	5 Experiments
		5.1 Experiment 1
		5.2 Experiment 2
	6 Conclusion
	References
Multiple Instance Learning for Unilateral Data
	1 Introduction
	2 The Proposed Method
		2.1 Bag Mapping
		2.2 NU Classification Based on Composite Mapping
		2.3 Analysis of Generalization Error Bounds
	3 Experiments
		3.1 Experiment Settings
		3.2 Results
	4 Conclusion
	References
An Online Learning Algorithm for Non-stationary Imbalanced Data by Extra-Charging Minority Class
	1 Introduction
	2 Background and Related Work
		2.1 Concept Drift
		2.2 Class Imbalance
	3 Proposed Method
	4 Experiments
		4.1 Competitors
		4.2 Datasets
		4.3 Performance Metrics
		4.4 Predictive Performance
	5 Model Behavior
		5.1 Nonstationary and Noisy Data
		5.2 Parameters
		5.3 Time and Space Complexity
	6 Conclusion
	References
Locally Linear Support Vector Machines for Imbalanced Data Classification
	1 Introduction
	2 Shortcomings of Global Approaches to Imbalanced Data
	3 LL-SVM for Imbalanced Data
		3.1 Training Procedure for LL-SVM
		3.2 Advantages of LL-SVM in Imbalanced Domain
	4 Experimental Study
		4.1 Experimental Set-Up
		4.2 Analysis of LL-SVM Neighborhood Parameter
		4.3 Results for Oversampling-Based Approaches
		4.4 Results for Cost-Sensitive Approaches
	5 Conclusions and Future Works
	References
Low-Dimensional Representation Learning from Imbalanced Data Streams
	1 Introduction
	2 Low-Dimensional Representation
	3 Low-Dimensional Projection for Imbalanced Data
	4 Experimental Study
		4.1 Data Stream Benchmarks
		4.2 Set-Up
		4.3 Experiment 1: Low-Dimensional Representations
		4.4 Experiment 2: Skew-Insensitive Algorithms
		4.5 Experiment 3: Evaluation on Real-World Data Streams
	5 Conclusions and Future Works
	References
PhotoStylist: Altering the Style of Photos Based on the Connotations of Texts
	1 Introduction
	2 Related Work
	3 Problem Statement
	4 Solution
		4.1 Image-Text ANP Crossmatcher (ITAC) Module
		4.2 ANP-Based Style Generator (ASG) Module
		4.3 Segmented Style Transfer (SST) Module
	5 Evaluation
		5.1 Baselines
		5.2 Crowdsourcing-Based Evaluation
		5.3 Visual Comparison
	6 Conclusion
	References
Gazetteer-Guided Keyphrase Generation from Research Papers
	1 Introduction
	2 Related Work
		2.1 Keyphrase Extraction
		2.2 Keyphrase Generation
	3 Problem Definition
	4 Baseline Architecture
	5 Gazetteer-Enhanced Architecture
		5.1 Automatic Gazetteer Construction
		5.2 Integrating Gazetteer Knowledge
	6 Experiments and Results
		6.1 Datasets
		6.2 Baselines and Evaluation Metrics
		6.3 Implementation
		6.4 Performance Comparison
		6.5 Analysis of the Number of Generated Keyphrases
		6.6 Case Study
	7 Conclusion
	References
Minits-AllOcc: An Efficient Algorithm for Mining Timed Sequential Patterns
	1 Introduction
	2 Problem Definitions
	3 Related Works
	4 The Proposed Algorithm: Minits-AllOcc
		4.1 Occurrence Tree (O-Tree)
		4.2 Overview
		4.3 The Proposed Algorithm: Minits-AllOcc
		4.4 The Proposed Enhancement
	5 Performance Analysis
		5.1 Experimental Setup
		5.2 Datasets and Experimental Parameters
		5.3 Competing Algorithms
		5.4 Evaluation Metrics
		5.5 Experimental Results
	6 Conclusion and Future Work
	References
T3N: Harnessing Text and Temporal Tree Network for Rumor Detection on Twitter
	1 Introduction
	2 Related Work
	3 T3N: A System for Rumor Detection on Twitter
		3.1 Problem Definition and T3N System Overview
		3.2 Text Encoder
		3.3 Tree Encoder
		3.4 Decayed Tree Encoder
		3.5 Temporal Tree Encoder
		3.6 Putting It All Together
	4 Experiments
		4.1 Datasets, Experimental Settings and Baselines
		4.2 Accuracy Comparison
		4.3 Ablation Studies
		4.4 Early Detection Results
	5 Conclusion
	References
AngryBERT: Joint Learning Target and Emotion for Hate Speech Detection
	1 Introduction
	2 Related Work
	3 Datasets and Tasks
		3.1 Primary Task and Datasets
		3.2 Secondary Tasks and Datasets.
	4 Proposed Model
		4.1 Problem Formulation
		4.2 Architecture of AngryBERT
		4.3 Training of AngryBERT
	5 Experiments
		5.1 Baselines
		5.2 Evaluation Metrics
		5.3 Experiment Results
		5.4 Ablation Study
		5.5 Case Studies
	6 Conclusion
	References
SCARLET: Explainable Attention Based Graph Neural Network for Fake News Spreader Prediction
	1 Introduction
	2 Interpersonal Trust and User Credibility Features
		2.1 Trust-Based Features
		2.2 Credibility-Based Features
	3 Proposed Approach
		3.1 Importance Score Using Attention:
		3.2 Feature Aggregation
		3.3 Node Classification
	4 Experimental Analysis
		4.1 Data Collection
		4.2 Analysis of F  T
		4.3 Models and Metrics
		4.4 Implementation Details
		4.5 Performance Evaluation
		4.6 Sensitivity Analysis
		4.7 Explainability Analysis of Trust and Credibility
	5 Conclusions and Future Work
	References
Content Matters: A GNN-Based Model Combined with Text Semantics for Social Network Cascade Prediction
	1 Introduction
	2 Related Work
		2.1 Information Cascade Prediction
		2.2 GNNs-Based Approaches
	3 Methods
		3.1 Problem Definition
		3.2 Topic Embedding Module
		3.3 The TSGNN Layer
		3.4 Gated Activation Unit
		3.5 Cascade Size Prediction
	4 Experiments
		4.1 Datasets
		4.2 Baselines
		4.3 Evaluation Metrics
	5 Results
		5.1 Performance Comparison
		5.2 Variants of TSGNN
		5.3 Parameter Analysis
		5.4 Ablation Experiment
	6 Conclusion
	References
TERMCast: Temporal Relation Modeling for Effective Urban Flow Forecasting
	1 Introduction
	2 Related Work
	3 Preliminaries
	4 Methodology
		4.1 Short-Term Prediction
		4.2 Long-Term Relation Prediction
		4.3 Prediction with Consistency
	5 Experiments
		5.1 Datasets and Metrics
		5.2 Implementation Details
		5.3 Comparison Against Other Methods
		5.4 Ablation Studies
		5.5 Different Weighted Fusion Configurations
	6 Conclusion
	References
Traffic Flow Driven Spatio-Temporal Graph Convolutional Network for Ride-Hailing Demand Forecasting
	1 Introduction
	2 Related Work
		2.1 Graph Convolution Network
		2.2 Deep Learning on Traffic Prediction
	3 Preliminaries
	4 Methodology
		4.1 Overview
		4.2 Traffic-Region Demand Graph Convolution Network Module for Spatial Modeling
		4.3 ConvLSTM for Temporal Correlation Modeling
		4.4 Demand Forecasting
	5 Experiments
		5.1 Datasets
		5.2 Experiment Settings
		5.3 Results
	6 Conclusion
	References
A Proximity Forest for Multivariate Time Series Classification
	1 Introduction
	2 Related Work
	3 Proximity Forest for Multivariate Time Series
		3.1 Construction of Forest
		3.2 Weighted Classifying
		3.3 Locally Slope-Based DTW
	4 Experiment and Evaluation
		4.1 Parameter Analysis
		4.2 Impact of Design Decisions
		4.3 Classification Accuracy
		4.4 Complexity Analysis
	5 Conclusion
	References
C2-Guard: A Cross-Correlation Gaining Framework for Urban Air Quality Prediction
	1 Introduction
	2 Related Work
	3 Problem Formulation
	4 The Proposed Framework
		4.1 Temporal Correlation Module
		4.2 Factor Correlation Module
		4.3 Cross Gaining Module
	5 Experiment
		5.1 Datasets
		5.2 Experimental Setup
		5.3 Performance Comparison
		5.4 Ablation Study
	6 Conclusion and Future Work
	References
Simultaneous Multiple POI Population Pattern Analysis System with HDP Mixture Regression
	1 Introduction
	2 Related Work
	3 Urban Population Pattern Analysis System with HDP Regression
	4 Definition of Proposed Method: HDP Mixture Regression
		4.1 Urban Dynamics Prediction by Bilinear Poisson Regression
		4.2 Definition of HDP Mixture Regression
		4.3 Prediction by HDP Mixture Regression
		4.4 Urban Dynamics Prediction Systems Using HDP Mixture Regression
	5 Experimental Results
		5.1 Dataset
		5.2 Evaluation Metric
		5.3 Comparison Methods
		5.4 Comparison with Previous Predictive Methods
		5.5 Application Using HDP Mixture Regression
	6 Conclusion
	References
Interpretable Feature Construction for Time Series Extrinsic Regression
	1 Introduction
	2 TSER via a Relational Way
	3 Experimental Validation
	4 Conclusion and Perspectives
	References
SEPC: Improving Joint Extraction of Entities and Relations by Strengthening Entity Pairs Connection
	1 Introduction
	2 Related Work
		2.1 Joint Extraction of Entities and Relations
		2.2 Dual Supervised Learning
	3 Methods and Technical Solutions
		3.1 Entity Tagger
		3.2 Entity Pair Recognizer: Strengthening Entity Pairs Connection
		3.3 Relation Tagger
	4 Empirical Evaluation
		4.1 Dataset and Evaluation Metrics
		4.2 Experimental Settings
		4.3 Implementation Detail
		4.4 Experimental Result Analysis
	5 Conclusion and Inspiration
	References
Author Index