Domain-Specific Knowledge Graph Construction (eBook)

Mayank Kejriwal (Autor)

eBook Download: PDF

2019 | 1st ed. 2019
XIV, 107 Seiten
Springer International Publishing (Verlag)
978-3-030-12375-8 (ISBN)

Infos

The vast amounts of ontologically unstructured information on the Web, including HTML, XML and JSON documents, natural language documents, tweets, blogs, markups, and even structured documents like CSV tables, all contain useful knowledge that can present a tremendous advantage to the Artificial Intelligence community if extracted robustly, efficiently and semi-automatically as knowledge graphs. Domain-specific Knowledge Graph Construction (KGC) is an active research area that has recently witnessed impressive advances due to machine learning techniques like deep neural networks and word embeddings. This book will synthesize Knowledge Graph Construction over Web Data in an engaging and accessible manner.

The book will describe a timely topic for both early -and mid-career researchers. Every year, more papers continue to be published on knowledge graph construction, especially for difficult Web domains. This work would serve as a useful reference, as well as an accessible but rigorous overview of this body of work. The book will present interdisciplinary connections when possible to engage researchers looking for new ideas or synergies. This will allow the book to be marketed in multiple venues and conferences. The book will also appeal to practitioners in industry and data scientists since it will have chapters on both data collection, as well as a chapter on querying and off-the-shelf implementations.

The author has, and continues to, present on this topic at large and important conferences. He plans to make the powerpoint he presents available as a supplement to the work.  This will draw a natural audience for the book. Some of the reviewers are unsure about his position in the community but that seems to be more a function of his age rather than his relative expertise. I agree with some of the reviewers that the title is a little complicated. I would recommend 'Domain Specific Knowledge Graphs'. 

Inhalt

Preface 7
Acknowledgments 9
Contents 10
Acronyms 12
1 What Is a Knowledge Graph? 14
1.1 Introduction 14
1.2 Example 1: Academic Domain 17
1.3 Example 2: Products and Companies 18
1.4 Example 3: Geopolitical Events 19
1.5 Conclusion 20
2 Information Extraction 21
2.1 Introduction 21
2.2 Challenges of IE 22
2.3 Scope of IE Tasks 23
2.3.1 Named Entity Recognition 24
2.3.1.1 Supervised Methods 25
2.3.1.2 Semi-supervised Methods 26
2.3.1.3 Unsupervised Methods 27
2.3.1.4 Features 28
2.3.2 Relation Extraction 34
2.3.3 Event Extraction 36
2.3.4 Web IE 38
2.4 Evaluating IE Performance 41
2.5 Summary 42
3 Entity Resolution 44
3.1 Introduction 44
3.2 Challenges and Requirements 45
3.3 Two-Step Framework 49
3.3.1 Blocking 50
3.3.1.1 Traditional Blocking 51
3.3.1.2 Sorted Neighborhood 52
3.3.1.3 Canopies 53
3.3.1.4 Research Frontier: Learning Blocking Keys 54
3.3.2 Similarity 55
3.4 Measuring Performance 58
3.4.1 Measuring Blocking Performance 59
3.4.2 Measuring Similarity Performance 61
3.5 Extending the Two-Step Workflow: A Brief Note 62
3.6 Related Work: A Brief Review 62
3.6.1 Automated ER Solutions 63
3.6.1.1 The Automation-Scalability Tradeoff 65
3.6.2 Structural Heterogeneity 66
3.6.3 Blocking Without Supervision: Where Do We Stand? 67
3.7 Summary 68
4 Advanced Topic: Knowledge Graph Completion 69
4.1 Introduction 69
4.2 Knowledge Graph Embeddings 71
4.2.1 TransE 73
4.2.2 TransE Extensions and Alternatives 74
4.2.3 Limitations and Alternatives 76
4.2.4 Research Frontiers and Recent Work 76
4.2.4.1 Ontological Information 77
4.2.4.2 Text 78
4.2.4.3 Other Extrinsic Information Sets 79
4.2.5 Applications of KGEs 82
4.3 Summary 84
5 Ecosystems 85
5.1 Introduction 85
5.2 Web of Linked Data 85
5.2.1 Linked Data Principles 87
5.2.2 Technology Stack 88
5.2.3 Linking Open Data 89
5.2.4 Example: DBpedia 90
5.3 Google Knowledge Vault 92
5.4 Schema.org 94
5.5 Where is the Future Going? 96
Glossary 98
References 101
Index 111

Produkt-Details

• Erscheint lt. Verlag: 4.3.2019
• Reihe/Serie: SpringerBriefs in Computer Science
• Reihe/Serie: SpringerBriefs in Computer Science
• Zusatzinfo: XIV, 107 p. 19 illus.
• Verlagsort: Cham
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Informatik
• Schlagworte: Data Mining • Domain Discovery • entity-centric search • entity resolution • information extraction • Knowledge Discory • Knowledge Graph Completion • Knowledge Graph Construction • Knowledge Graph Embeddings • knowledge graphs • machine learning • Natural Language Processing • Probabilistic Soft Logic • Querying • semantic web • Web Corpora • Wrapper Induction
• ISBN-10: 3-030-12375-8 / 3030123758
• ISBN-13: 978-3-030-12375-8 / 9783030123758