Entity Alignment : : Concepts, Recent Advances and Novel Approaches.

• Superior document: Big Data Management Series
• Author: Zhao, Xiang.
• TeilnehmendeR: Zeng, Weixin.; Tang, Jiuyang.
• Place / Publishing House: Singapore : : Springer Singapore Pte. Limited,, 2023.; Ã2023.
• Year of Publication: 2023
• Edition: 1st ed.
• Language: English
• Series: Big Data Management Series
• Physical Description: 1 online resource (252 pages)

Table of Contents

• Intro
• Preface
• Background
• Content Organization
• Contents
• Part I Concept and Categorization
• 1 Introduction to Entity Alignment
• 1.1 Background
• 1.2 Related Works
• 1.2.1 Entity Linking
• 1.2.2 Entity Resolution
• 1.2.3 Entity Resolution on KGs
• 1.3 Evaluation Settings
• References
• 2 State-of-the-Art Approaches
• 2.1 Introduction
• 2.2 A General EA Framework
• 2.2.1 Embedding Learning Module
• 2.2.2 Alignment Module
• 2.2.3 Prediction Module
• 2.2.4 Extra Information Module
• 2.3 Experiments and Analysis
• 2.3.1 Categorization
• 2.3.2 Experimental Settings
• 2.3.3 Results and Analyses on DBP15K
• 2.3.4 Results and Analyses on SRPRS
• 2.3.5 Results and Analyses on DWY100K
• 2.3.6 Efficiency Analysis
• 2.3.7 Comparison with Unsupervised Approaches
• 2.3.8 Module-Level Evaluation
• 2.3.9 Summary
• 2.3.10 Guidelines and Suggestions
• 2.4 New Dataset and Further Experiments
• 2.4.1 Dataset Construction
• 2.4.2 Experimental Results on DBP-FB
• 2.4.3 Unmatchable Entities
• 2.5 Conclusion
• Appendix A
• Methods in Group i of Table 2.1
• Methods in Group ii of Table 2.1
• Methods in Group iii of Table 2.1
• Appendix B
• Parameter Setting
• References
• Part II Recent Advances
• 3 Recent Advance of Representation Learning Stage
• 3.1 Overview
• 3.2 Models
• 3.2.1 ALiNet
• 3.2.2 MRAEA
• 3.2.3 RREA
• 3.2.4 RPR-RHGT
• 3.2.5 RAGA
• 3.2.6 Dual-AMN
• 3.2.7 ERMC
• 3.2.8 KE-GCN
• 3.2.9 RePS
• 3.2.10 SDEA
• 3.3 Experiments
• 3.3.1 Experimental Setting
• 3.3.2 Overall Results and Analysis
• 3.3.3 Further Experiments
• 3.3.3.1 Pre-processing Module
• 3.3.3.2 Messaging Module
• 3.3.3.3 Attention Module
• 3.3.3.4 Aggregation Module
• 3.3.3.5 Post-processing Module
• 3.3.3.6 Loss Function Module
• 3.4 Conclusion
• References
• 4 Recent Advance of Alignment Inference Stage.
• 4.1 Introduction
• 4.2 Preliminaries
• 4.2.1 Task Formulation and Framework
• 4.2.2 Related Work and Scope
• 4.2.3 Key Assumptions
• 4.3 Alignment Inference Algorithms
• 4.3.1 Overview
• 4.3.2 Simple Embedding Matching
• 4.3.3 CSLS Algorithm
• 4.3.4 Reciprocal Embedding Matching
• 4.3.5 Embedding Matching as Assignment
• 4.3.6 Stable Embedding Matching
• 4.3.7 RL-Based Embedding Matching
• 4.4 Main Experiments
• 4.4.1 EntMatcher: An Open-Source Library
• 4.4.2 Experimental Settings
• 4.4.3 Main Results and Comparison
• 4.4.4 Results on Large-Scale Datasets
• 4.4.5 Analysis and Insights
• 4.5 New Evaluation Settings
• 4.5.1 Unmatchable Entities
• 4.5.2 Non-1-to-1 Alignment
• 4.6 Summary and Future Direction
• 4.7 Conclusion
• References
• Part III Novel Approaches
• 5 Large-Scale Entity Alignment
• 5.1 Introduction
• 5.2 Framework
• 5.3 Partition Strategies for Entity Alignment
• 5.3.1 Seed-Oriented Unidirectional Graph Partition
• 5.3.2 Bidirectional Graph Partition
• 5.3.3 Iterative Bidirectional Graph Partition
• 5.3.4 Complexity Analysis
• 5.3.5 Discussion
• 5.4 Reciprocal Alignment Inference
• 5.4.1 Entity Structural Representation Learning
• 5.4.2 Preference Modeling
• 5.4.3 Preference Aggregation
• 5.4.4 Correctness Analysis
• 5.4.5 Complexity Analysis
• 5.5 Variants of Reciprocal Alignment Inference
• 5.5.1 No-Ranking Aggregation
• 5.5.2 Progressive Blocking
• 5.6 Experimental Settings
• 5.6.1 Dataset
• 5.6.2 Construction of a Large-Scale Dataset
• 5.6.3 Implementation Details
• 5.6.4 Evaluation Metrics
• 5.6.5 Competing Methods
• 5.7 Results
• 5.7.1 Evaluation on Large-Scale Dataset
• 5.7.2 Comparison with State-of-the-Art Methods
• 5.7.3 Experiments and Analyses on Partitioning
• 5.7.4 Experiments and Analyses on Reciprocal Inference.
• 5.7.5 Experiments and Analyses on Progressive Blocking
• 5.8 Related Work
• 5.9 Conclusion
• Appendix
• Correctness Analysis
• References
• 6 Long-Tail Entity Alignment
• 6.1 Introduction
• 6.2 Related Work
• 6.3 Impact of Long-Tail Phenomenon
• 6.4 Methodology
• 6.4.1 Name Representation Learning
• 6.4.2 Degree-Aware Co-attention Feature Fusion
• 6.4.3 Iterative KG Completion
• 6.5 Experiments
• 6.5.1 Experimental Setting
• 6.5.2 Results
• 6.5.3 Ablation Study
• 6.5.4 Error Analysis
• 6.5.5 Further Experiment
• 6.6 Conclusion
• References
• 7 Weakly Supervised Entity Alignment
• 7.1 Introduction
• 7.2 Preliminaries
• 7.2.1 Problem Formulation
• 7.2.2 Model Overview
• 7.3 Reinforced Active Learning
• 7.3.1 Query Strategies
• 7.3.2 Reinforced Active Entity Selection via MAB
• 7.4 Contrastive Embedding Learning
• 7.4.1 Semi-supervised Alignment Loss
• 7.4.2 Graph Encoders
• 7.4.3 Unsupervised Contrastive Loss
• 7.4.4 Alignment Inference
• 7.5 Experiment
• 7.5.1 Experimental Settings
• 7.5.2 Main Results (RQ1)
• 7.5.3 Experiments on Contrastive Learning (RQ2)
• 7.5.4 Experiments on Reinforced AL (RQ3)
• 7.6 Related Work
• 7.7 Conclusion
• References
• 8 Unsupervised Entity Alignment
• 8.1 Introduction
• 8.2 Task Definition and Related Work
• 8.3 Methodology
• 8.3.1 Model Outline
• 8.3.2 Side Information
• 8.3.3 Unmatchable Entity Prediction
• 8.3.3.1 Thresholded Bidirectional Nearest Neighbor Search
• 8.3.3.2 Confidence-Based TBNNS
• 8.3.4 The Progressive Learning Framework
• 8.3.4.1 Knowledge Graph Representation Learning
• 8.3.4.2 The Progressive Learning Algorithm
• 8.3.4.3 Dynamic Threshold Adjustment
• 8.4 Experiment
• 8.4.1 Experimental Settings
• 8.4.2 Results
• 8.4.2.1 Results Using Low-Quality Side Information
• 8.4.3 Ablation Study
• 8.4.4 Quantitative Analysis
• 8.5 Conclusion.
• References
• 9 Multimodal Entity Alignment
• 9.1 Introduction
• 9.2 Related Work
• 9.2.1 Multi-Modal Knowledge Graph
• 9.2.2 Representation Learning in Hyperbolic Space
• 9.3 Preliminaries
• 9.3.1 Task Formulation
• 9.3.2 Graph Convolutional Neural Networks
• 9.3.3 Hyperboloid Manifold
• 9.4 Methodology
• 9.4.1 Structural Representation Learning
• 9.4.2 Visual Representation Learning
• 9.4.3 Multi-Modal Information Fusion
• 9.4.4 Alignment Prediction
• 9.4.5 Model Training
• 9.5 Experiment
• 9.5.1 Dataset and Evaluation Metric
• 9.5.2 Experimental Setting and Competing Approaches
• 9.5.3 Results
• 9.5.4 Ablation Experiment
• 9.5.5 Case Study
• 9.5.6 Additional Experiment
• 9.6 Conclusion
• References.