Database Systems for Advanced Applications : 28th International Conference, DASFAA 2023, Tianjin, China, April 17–20, 2023, Proceedings, Part III

The four-volume set LNCS 13943, 13944, 13945 and 13946 constitutes the proceedings of the 28th International Conference on Database Systems for Advanced Applications, DASFAA 2023, held in April 2023 in Tianjin, China. The total of 125 full papers, along with 66 short papers, are presented together in this four-volume set was carefully reviewed and selected from 652 submissions. Additionally, 15 industrial papers, 15 demo papers and 4 PhD consortium papers are included. The conference presents papers on subjects such as model, graph, learning, performance, knowledge, time, recommendation, representation, attention, prediction, and network. Preface Organization Contents – Part III Graph and Networks Meta-Path Based Social Relation Reasoning in a Deep and Robust Way 1 Introduction 2 Task and Related Works 2.1 Social Relation Reasoning 2.2 Related Works 3 Heterogeneous Graph Variational Autoencoders 3.1 Heterogeneous Graph Encoder 3.2 Multi-Signal Decoder 3.3 HGVAE Joint Training 4 Experiments 4.1 Dataset 4.2 Baseline Methods 4.3 Implementation Details 4.4 Experiment Results 4.5 Ablation Studies 4.6 In-depth Model Analysis 4.7 Case Study 5 Conclusion References SRACas: A Social Role-Aware Graph Neural Network-Based Model for Popularity Prediction of Information Cascades 1 Introduction 2 Preliminaries 3 Proposed Method 3.1 Local Structure Learning 3.2 Social Role-Aware Attention 3.3 Temporal Feature Learning 3.4 Prediction and Loss Function 3.5 Sparse Encoding 4 Experiments 4.1 Experimental Setups 4.2 Prediction Performance 4.3 Visualization and Explanation 5 Conclusion References Few-Shot Link Prediction for Event-Based Social Networks via Meta-learning 1 Introduction 2 Related Works 3 Problem Definition 4 Methodology 4.1 Cross-Network Task Sampling 4.2 Event-Aware Link Prediction Task 4.3 Meta-learning Framework for EBSNs 5 Experiments 5.1 Experiment Setup 5.2 Experiments Results 6 Conclusion References Mining Discriminative Sub-network Pairs in Multi-frequency Brain Functional Networks 1 Introduction 2 Methodology 2.1 Multi-frequency Brain Functional Network Construction 2.2 Frequent Subgraph Mining 2.3 Discriminative Subgraph Pair (DSP) Mining 2.4 Feature Selection and Classification 3 Result and Discussion 3.1 Experimental Settings 3.2 Result Analysis 3.3 Influence of Parameters 4 Related Work 5 Conclusions References Local Spectral for Polarized Communities Search in Attributed Signed Network 1 Introduction 2 Related Work 3 Preliminaries 4 Method 4.1 Construction of Augmented Signed Graph 4.2 Choice of Combination Factor 4.3 Local Spectral Subspace 4.4 Discussion on Quality of Polarized Communities 5 Experiments 5.1 Datasets Description 5.2 Experimental Settings 5.3 Performance Evaluation 5.4 Parameter Sensitivity Analysis 5.5 Case Study 6 Conclusion and Future Work References Subgraph Reconstruction via Reversible Subgraph Embedding 1 Introduction 1.1 Motivation 1.2 Our Approach and Contributions 2 Problem Definition and Framework 2.1 Problem Formulation 2.2 Compressed Sensing 2.3 Framework 3 Reversible Subgraph Embedding 3.1 Node Embedding 3.2 Subgraph Embedding 4 Subgraph Reconstruction 5 Experiments 5.1 Experiment Setup 5.2 Subgraph Reconstruction 5.3 Graph Reconstruction 5.4 Node Classification 5.5 Graph Classification 5.6 Discussion 6 Related Work 7 Conclusion References Efficiently Answering Why-Not Questions on Radius-Bounded k-Core Searches 1 Introduction 2 Related Work and Problem Statement 2.1 Related Work 2.2 Preliminary 2.3 Problem Definition 3 Influence Factors Analysis 3.1 The Analysis of k 3.2 The Analysis of r 4 Explanation Algorithms 4.1 Modifying the Cohesiveness Constraint k 4.2 Modifying the Spatial Constraints r 5 Experiments 5.1 Experimental Setup 5.2 Performance Evaluation 6 Conclusion References Multi-scale Community Detection in Subspace of Attribute 1 Introduction 2 Preliminaries 3 Method 3.1 Scale-dependent Node Representation 3.2 Community Detection with Attribute Subspaces 4 Experiments 4.1 Experimental Setup 4.2 Performance Comparison 5 Conclusion References Efficient Anomaly Detection in Property Graphs 1 Introduction 2 Problem Definition 3 ACGPMiner Approach 3.1 Overview 3.2 Detail Design 3.3 Discussion 4 Optimizations 4.1 Pre-search Pruning 4.2 Load Balancing 5 Evaluation 5.1 Experimental Setup 5.2 Experimental Results 5.3 Optimization Analysis 6 Related Work 7 Conclusion References Edge Coloring on Dynamic Graphs 1 Introduction 2 Related Work 3 Preliminaries 4 Our Approach 4.1 General Idea and Problem Analysis 4.2 A Basic Algorithm 4.3 An Optimized Approach 4.4 Early Pruning 5 Performance Studies 6 Conclusion References Discovering Persistent Subgraph Patterns over Streaming Graphs 1 Introduction 2 Problem Formulation 3 The Baseline Solution 4 The fastPP Framework 4.1 The Fast Algorithm fastPP 5 TFD +: An Optimized Auxiliary Data Structure 6 Experiments 6.1 Performance Evaluation 6.2 Case Study 7 Related Work 8 Conclusion References MRSCN: A GNN-based Model for Mining Relationship Strength Changes Between Nodes in Dynamic Networks 1 Introduction 2 Related Work 3 Preliminaries 4 Mining Relationship Strength Changes Between Nodes 4.1 Global Structure Information Capture 4.2 Graph Neural Network Model 4.3 Relationship Strength Change Computation 5 Drastic Group Mining 6 Experiments 6.1 Experimental Setup 6.2 Experimental Results 7 Conclusion References TE-DyGE: Temporal Evolution-Enhanced Dynamic Graph Embedding Network 1 Introduction 2 Related Works 2.1 Static Graph Representation Learning 2.2 Dynamic Graph Representation Learning 3 Problem Definition 4 Framework 4.1 Fine-grained Temporal Weighted Attention 4.2 Coarse-grained Temporal Attention Module 4.3 Objective Function 5 Experiments 5.1 Datasets 5.2 Baselines 5.3 Experimental Setup 5.4 Effectiveness of TE-DyGE 5.5 Effectiveness of Proposed Contributions 5.6 Parameters Sensitivity 6 Conclusion References Mining Top-k Frequent Patterns over Streaming Graphs 1 Introduction 2 Preliminaries 3 The Baseline Solution 4 TopKF: A Progressive Solution 4.1 Problem Analysis 4.2 The Progressive Algorithm Framework 4.3 Mathematical Analysis 5 FPCS: Augmented Auxiliary Data Structure 6 Experiments 6.1 Experiments on Different Metrics 6.2 Experiments on Varying Parameters 7 Related Work 8 Conclusion References An Efficient Index-Based Method for Skyline Path Query over Temporal Graphs with Labels 1 Introduction 2 Related Work 3 Problem Definition 4 MP-Index Construction 4.1 Main Points Discovering 4.2 Mout Set Recording 5 Skyline Path Query Based on MP Index 5.1 Bidirectional Topology 5.2 Query Algorithm Based on MP-Index 6 Experiment 6.1 Experimental Settings 6.2 Index Evaluation 6.3 Query Time Evaluation and k Value Influence 6.4 Influence of the Number of Labels 7 Conclusion References Efficient and Scalable Distributed Graph Structural Clustering at Billion Scale 1 Introduction 2 Preliminary 3 Related Works 4 Our Approach 4.1 A Fine-grained Framework for Clustering 4.2 Architecture 4.3 Algorithm Implementation 4.4 Load Balance 4.5 Algorithm Analysis 5 Evaluation 6 Conclusion References Hierarchical All-Pairs SimRank Calculation 1 Introduction 2 Preliminaries 2.1 Background of SimRank 2.2 Linear System for SimRank 2.3 Numerical Solution 3 A Hierarchical All-Pairs SimRank Calculation Framework 3.1 Hierarchical Framework 3.2 Optimization 4 Related Work 4.1 All-Pairs SimRank Algorithms 4.2 Other SimRank Algorithms 5 Experiments 6 Conclusions References Contraction Hierarchies with Label Restrictions Maintenance in Dynamic Road Networks 1 Introduction 2 Related Work 3 Preliminaries 3.1 Contraction Hierarchies with Label Restrictions 3.2 Problem Statement 4 Baseline Approach 5 CHLR Maintenance Algorithm 5.1 Optimization Strategy 6 Performance Studies 7 Conclusion References GRMI: Graph Representation Learning of Multimodal Data with Incompleteness 1 Introduction 2 Problem Formulation 3 Methodology 3.1 GNN Architecture 3.2 Embeddings for Modal Imputation and Modal Fusion 3.3 Self-supervised Learning Strategy 4 Experiment 4.1 Data 4.2 Baseline Methods 4.3 Experimental Setup 4.4 Result Analysis 5 Conclusion References Efficient Network Representation Learning via Cluster Similarity 1 Introduction 2 Preliminaries 3 Proposed Method 3.1 Similarity Between Clusters 3.2 Dimensionality Expansion 3.3 SVD Computation 3.4 Representation Learning Algorithm 4 Experimental Evaluation 4.1 Network Representation Learning Time 4.2 Multi-label Node Classification 5 Conclusions References Learning with Small Data: Subgraph Counting Queries 1 Introduction 2 Preliminaries 2.1 GNN-Based Encoder for Subgraph Counting 3 A Meta Learning Approach 3.1 RGIN Gaussian Process (RGIN-GP) 3.2 Meta Learning for RGIN-GP 3.3 Feature Encoding 4 Experimental Studies 5 Conclusion References FairHELP: Fairness-Aware Heterogeneous Information Network Embedding for Link Prediction 1 Introduction 2 Related Works 2.1 Heterogeneous Information Network Embedding 2.2 Fairness in Machine Learning 3 Preliminaries 4 Methods 4.1 Network Embedding Generator 4.2 Link Predictor 4.3 Semantic Subgroup Discriminator 4.4 Model Training 5 Experiments 5.1 Experimental Settings 5.2 Bias Mitigation Performance in Link Prediction 5.3 Parameter Analysis 6 Conclusion References HAEP: Heterogeneous Environment Aware Edge Partitioning for Power-Law Graphs 1 Introduction 2 Related Work 3 Problem Definition 4 HAEP 4.1 Neighbor Expansion 4.2 Center Boundary Vertices 4.3 Distributed HAEP 5 Experiments 5.1 Experiment Settings 5.2 Experiment Results 6 Conclusion References A Graph Embedding Approach for Link Prediction via Triadic Closure Based Direct Aggregation and Weighted Concatenation 1 Introduction 2 Related Work 3 Method 3.1 Triadic Closure Based Direct Aggregation 3.2 Weighted Concatenation for Edge Embedding 3.3 Link Prediction 4 Experiments 4.1 Datasets and Settings 4.2 Performance of Link Prediction 4.3 Performance of Node Direct Aggregation 4.4 Performance of Edge Concatenation 5 Conclusion References MPGCL: Multi-perspective Graph Contrastive Learning 1 Introduction 2 Related Work 3 Method 3.1 Problem Formulation and Graph Neural Network 3.2 Overall Framework 3.3 Graph Data Augmentation 3.4 Network Structure 3.5 Cross-View Contrastive Learning 4 Experiment 4.1 Dataset 4.2 Baselines 4.3 Experimental Setup 4.4 Result and Analysis 4.5 Ablation Study 4.6 Analysis of Hyper-parameters 4.7 Visualization 5 Conclusion References Retrieval A Joint Link-Retrieve Framework for Open Table-and-Text Question Answering 1 Introduction 2 Related Work 2.1 Open Domain Question Answering 2.2 Open Table and Text Question Answering 3 Methodology 3.1 Linking Table Blocks with Passages 3.2 Retrieving Evidence Candidates 3.3 Table-to-Text Generation 3.4 Reasoning and Answering 4 Experiments 4.1 Experiment Setup 4.2 Dataset and Evaluation 4.3 Baseline Methods 4.4 Main Results 4.5 Ablation Study 5 Conclusion References L2QA: Long Legal Article Question Answering with Cascaded Key Segment Learning 1 Introduction 2 Methodology 2.1 Sifter: Answer-Guided Key Segment Selection Module 2.2 Reader: Semantic Representation Module 2.3 Responder: Answer Predictor 3 Experiment 3.1 Dataset and Experimental Setup 3.2 Main Results 3.3 Ablation Study 4 Conclusions References An Adaptive Video Clip Sampling Approach for Enhancing Query-Based Moment Retrieval in Videos*-12pt 1 Introduction 2 Analysis of the Existing Fixed Size Sampling Methods 2.1 Problem Formulation 2.2 Problems of Existing Video Representations with Fixed Sampling Size 3 An Adaptive Sampling 3.1 Framework Overview 3.2 Backbone Network 3.3 Resampling Missing Clips 3.4 Enhancing Sparse Sampled Clips 3.5 Consistency Loss Maintenance 4 Experiments 4.1 Datasets and Evaluation Metrics 4.2 Performance Comparisons 5 Conclusion References Video Retrieval with Tree-Based Video Segmentation*-12pt 1 Introduction 2 Related Work 2.1 Video and Language Understanding for T2V Retrieval 2.2 Multiple Choice Learning 3 Proposed Methods 3.1 Tree-Based Video Segmentation 3.2 Global-Local Video Representation 3.3 MCL-CLIP4Clip 4 Experiments 4.1 Datasets 4.2 Experimental Details 4.3 Experimental Results 5 Conclusion References CMT: Cross-modal Memory Transformer for Medical Image Report Generation 1 Introduction 2 Cross-modal Memory Transformer 2.1 Visual Encoder 2.2 Medical Term Enhanced Module 2.3 Cross-modal Feature Memory Decoder 2.4 Multi-modal Feature Fusion Module 2.5 Training 3 Experiment 3.1 Experimental Settings 3.2 Results on Report Generation 3.3 Ablation Studies 4 Conclusion References Fintech Key-Phrase: A New Chinese Financial High-Tech Dataset Accelerating Expression-Level Information Retrieval 1 Introduction 2 Related Work 2.1 Expression-Level Information Extraction 2.2 Domain-Specific Benchmarks 3 Fintech Key-Phrase: A New Dataset for Expression-Level Information Retrieval 3.1 Main Motivation 3.2 Dataset Construction Guidelines 3.3 Dataset Statistics 4 Experiments 4.1 Baseline Models 4.2 Performance Results 5 Case Study 6 Released Tools for Financial High-Tech Domain Information Retrieval 6.1 Website of Our Fintech Key-Phrase 6.2 Released APIs 7 Conclusion References BACH: Black-Box Attacking on Deep Cross-Modal Hamming Retrieval Models*-12pt 1 Introduction 2 Background 2.1 Deep Cross-Modal Retrieval and Problem Formulation 3 Black-Box Attack on DCMHR Models 3.1 Black-Box Attack Framework 4 Experiment 4.1 Dataset 4.2 Evaluation 5 Related Work 5.1 Deep Cross-Modal Hashing 5.2 Adversarial Attacks 6 Conclusion References Category-Highlighting Transformer Network for Question Retrieval*-12pt 1 Introduction 2 Preliminaries 2.1 Related Work of the Identification Stage 3 Category-Highlighting Transformer Network 3.1 Category Identification Unit 3.2 Category-Highlighting Transformer 4 Experiments 4.1 Experimental Setup 4.2 Experimental Results 5 Conclusion References Text Processing HanoiT: Enhancing Context-aware Translation via Selective Context 1 Introduction 2 Our Approach 2.1 Problem Statement 2.2 HanoiT 2.3 Bi-lingual Context Integration 2.4 Training 3 Experiments 3.1 Datasets 3.2 Implementation Details 3.3 Baselines 3.4 Main Results 4 Analysis 5 Related Work 6 Conclusion References Speculation and Negation Scope Resolution via Machine Reading Comprehension Formulation with Data Augmentation*-12pt 1 Introduction 2 Approach 3 Experimentation 3.1 Experimental Settings 3.2 Overall Results 3.3 Detailed Analysis 4 Related Work 5 Conclusion References CoDE: Contrastive Learning Method for Document-Level Event Factuality Identification*-12pt 1 Introduction 2 Related Work 2.1 Event Factuality Identification 2.2 Contrastive Learning 3 Methodology 3.1 Task Definition 3.2 Overview 3.3 Document Encoding 3.4 Implicit Graph-Level Optimization 3.5 Contrastive Learning Objective 4 Experimentation 4.1 Experimental Settings 4.2 Baselines 4.3 Results and Analysis 4.4 Ablation Study 4.5 Case Study 4.6 Error Analysis 5 Conclusion References Recovering Missing Key Information: An Aspect-Guided Generator for Abstractive Multi-document Summarization 1 Introduction 2 Related Work 3 Methodology 4 Experiments 4.1 Dataset 4.2 Baselines 4.3 Implementation Details 4.4 Main Results 4.5 Case Study 5 Conclusion References TETA: Text-Enhanced Tabular Data Annotation with Multi-task Graph Convolutional Network 1 Introduction 2 Problem Definition 3 TETA Architecture 3.1 Text Extraction 3.2 Graph Construction 3.3 Representation Learning 3.4 Multi-task Learning 4 Experiments 4.1 Experiment Setup 4.2 Experimental Results 5 Related Work 6 Conclusion References A Two-Stage Label Rectification Framework for Noisy Event Extraction 1 Introduction 2 Methodology 2.1 Problem Statements 2.2 Framework Overview 2.3 Event Schema Mapping Stage 2.4 Self-adaptive Iteration Stage 2.5 Cooperative Global Pointer Network 3 Experiments 3.1 Experimental Setup 3.2 Main Results 3.3 Ablation Study 4 Conclusion References A Unified Visual Prompt Tuning Framework with Mixture-of-Experts for Multimodal Information Extraction 1 Introduction 2 Overview 3 Method 3.1 Diverse Image Encoders 3.2 Visual Prompts Fusion Module 3.3 Visual-Enhanced Text Encoder 3.4 Task-Specific Decoder 4 Experiment 4.1 Dataset 4.2 Parameter Settings 4.3 Baselines 4.4 Performance Comparison 4.5 Ablation Study 5 Conclusion References Is a Single Embedding Sufficient? Resolving Polysemy of Words from the Perspective of Markov Decision Process 1 Introduction 2 Related Work 3 Modeling RNN as Markov Decision Process 3.1 Markov Decision Process 3.2 MDP with a Definite Policy Function 3.3 Single-State Recurrent Neural Network 4 Polymorphic Recurrent Neural Network 4.1 Modeling PRNN as MDP 4.2 Network Structure and Policy Function 5 Experiments 5.1 Experimental Parameters 5.2 Experimental Results and Analysis 5.3 Stability of PRNN 6 Conclusions and Future Work References Unleashing Pre-trained Masked Language Model Knowledge for Label Signal Guided Event Detection 1 Introduction 2 Methodology 2.1 Trigger Augmentation 2.2 Label Signal Guided Event Classification 2.3 Training 3 Experiments 3.1 Settings 3.2 Overall Performance 3.3 Ablation Study 3.4 Parameter Analysis 4 Conclusions References Wukong-CMNER: A Large-Scale Chinese Multimodal NER Dataset with Images Modality 1 Introduction 2 Related Work 2.1 Chinese NER 2.2 Multimodal NER 3 Dataset Acquisition and Comparison 3.1 Dataset Collection and Annotation 3.2 Dataset Comparison 4 Methodology 4.1 Overview 4.2 Lexicon-Based Prompting Visual Clue Extraction Module 4.3 Cross-Modal Alignment Module 4.4 Cross-Modal Fusion 4.5 Model Training 5 Experiments 5.1 Experimental Setups 5.2 Main Results 5.3 Ablation Study 6 Conclusion References CAB: Empathetic Dialogue Generation with Cognition, Affection and Behavior 1 Introduction 2 Related Work 3 Method 3.1 Task Formulation and Overview 3.2 Emotional Context Encoder 3.3 Prior Network and Recognition Network (Affection) 3.4 Knowledge Acquisition and Fusion (Congnition) 3.5 Dialogue Act Predictor and Representation (Behavior) 3.6 Response Generation 3.7 Training Objectives 4 Experiments 4.1 Experimental Setup 4.2 Results and Analysis 5 Conclusions References Multimodal Entity Linking with Mixed Fusion Mechanism 1 Introduction 2 Related Work 3 Methodology 3.1 Preliminary 3.2 Overview of Mert-MEL 3.3 Input Design 3.4 Preprocess 3.5 Encoders 3.6 Multi-level Feature Extraction 3.7 Global Fusion 3.8 Bottleneck Fusion 3.9 Contrastive Learning 3.10 MEL Head 4 Experiments 4.1 Datasets, Baselines and Settings 4.2 Main Experimental Results 4.3 Ablation Experiments 4.4 Parameter Sensitivity Analysis 4.5 Case Study 5 Conclusion and Future Work References Optimizing Empathetic Response by Generating and Integrating Emotion Feedback and Topic Discussion 1 Introduction 2 Related Work 3 Methology 3.1 Context Encoding 3.2 Sub-response Generation 3.3 Sub-response Integration 3.4 Model Training 4 Experimental Setup 4.1 Dataset 4.2 Comparison Models 4.3 Implementation Details 4.4 Evaluation Metrics 5 Results and Discussions 5.1 Response-Generation Performance 5.2 Ablation Study 5.3 Case Study 5.4 Error Analysis and Outlook 6 Conclusion References Improving Event Representation with Supervision from Available Semantic Resources 1 Introduction 2 Related Work 3 Methodology 4 Experiments 4.1 Dataset and Implementation Details 4.2 Performance on Event Similarity Tasks 4.3 Performance on Script Prediction Task 4.4 Ablation Study 4.5 Training Efficiency 5 Conclusion References Select, Extend, and Generate: Generative Knowledge Selection for Open-Domain Dialogue Response Generation 1 Introduction 2 Methodology 2.1 Preliminary 2.2 Problem Definition and Overview 2.3 Generative Knowledge Selection 2.4 Dialogue Response Generation 2.5 Training 3 Experiment 3.1 Settings 3.2 Automatic Evaluation 3.3 Human Evaluation 3.4 More Analyses 3.5 Case Study 4 Related Work 5 Conclusion References Unify the Usage of Lexicon in Chinese Named Entity Recognition 1 Introduction 2 Related Work 3 Formalization of CNER Task 3.1 Sequence Labeling 3.2 Character Relation Classification 4 Proposed Method 4.1 Observation 4.2 Formalization of LWI Task 4.3 Network Architecture 4.4 Training Paradigm 5 Experiments 5.1 Experimental Setup 5.2 Effectiveness Study 5.3 Flexibility Study 5.4 Influence of Sparsity Phenomena 5.5 Ablation Study 6 Conclusion References A Prompt-Based Representation Individual Enhancement Method for Chinese Idiom Reading Comprehension 1 Introduction 2 Related Work 3 PRIEM Model 3.1 Model Frame 3.2 Learning Idioms Fusion Representation 3.3 Learning Idioms Individual Representation 3.4 Selecting the Correct Candidate Idiom 3.5 Method Integration 4 Experiments 4.1 Experimental Setup 4.2 Experimental Results and Discussion 4.3 Ablation Study 4.4 Joint Projection Parameter Adjustment 5 Conclusion References Cross-Modal Contrastive Learning for Event Extraction 1 Introduction 2 Model 2.1 Task Definition 2.2 Text Event Extractor 2.3 Video Event Extractor 2.4 Cross-Modal Contrastive Learner 2.5 Training and Inference 3 TVEE Dataset 3.1 Data Collection 3.2 Data Annotation 4 Experiments 4.1 Datasets 4.2 Evaluation Metrics 4.3 Compared Methods 4.4 Implementation Details 4.5 Main Results 4.6 Ablation Study 5 Related Work 5.1 Event Extraction 5.2 Contrastive Learning 6 Conclusion References Distinguishing Sensitive and Insensitive Options for the Winograd Schema Challenge 1 Introduction 2 Related Work 3 Our Approach 3.1 Option Weakening 3.2 Intermediate-Task Transfer Learning 4 Experiments 4.1 Implementation Details 4.2 Analysis of the Dataset 4.3 Overall Results 4.4 Ablation Study 5 Conclusion References Semi-supervised Learning for Fine-Grained Entity Typing with Mixed Label Smoothing and Pseudo Labeling*-12pt 1 Introduction 2 Overview 2.1 Problem Definition 2.2 Framework 3 Method 3.1 Mixed Label Smoothing for Labeled Data 3.2 Pseudo Labeling for Unlabeled Data 3.3 Training Process 4 Experiment 4.1 Datasets and Metrics 4.2 Baselines 4.3 Performance Comparison 4.4 Ablation Study 5 Conclusion References Meta-learning Siamese Network for Few-Shot Text Classification*-12pt 1 Introduction 2 Related Work 2.1 Meta-learning 2.2 Few-Shot Text Classification 3 Background 4 Algorithm 4.1 Word Representation Layer 4.2 Task Sampler 4.3 Constructing Sample Pairs 4.4 Weight Generator 4.5 Siamese Network 5 Experiments 5.1 Datasets 5.2 Experiment Setup 5.3 Classification Results 5.4 Ablation Study 5.5 Visualization 5.6 Hyper-parameter Sensitivity Analysis 6 Conclusion References Author Index