Linked Data (eBook)

Sherif Sakr is a professor of computer and information science in the Health Informatics department at King Saud bin Abdulaziz University for Health Sciences, and is also affiliated with the University of New South Wales and DATA61/CSIRO in Australia. Sherif’s research interests revolve around the areas of efficient and scalable Big Data Management, Processing and Analytics. In 2013, he was awarded the Stanford Innovation and Entrepreneurship Certificate.Marcin Wylot is a postdoctoral researcher at TU Berlin, Germany, in the Open Distributed Systems group. His main research interests are in database systems for Semantic Web data, provenance in Linked Data, Internet of Things, and Big Data processing.Raghava Mutharaju is a research scientist in the AI & Machine Learning Systems division of GE Global Research in Niskayuna, NY, USA. His research interests are in ontology modeling and reasoning, scalable SPARQL query processing, Big Data, Semantic Web and its applications.Danh Le Phuoc is a Marie Sklodowaka-Curie Fellow at TU Berlin. He is working on Pervasive Analytics which includes Linked Data/Semantic Web, Pervasive Computing, Future Internet and Big Data for Internet of Everything.Irini Fundulaki is a Principal Researcher at the Institute of Computer Science of the Foundation for Research and Technology-Hellas. Her research interests are related to Web Data Management and more specifically the development of benchmarks for RDF engines, instance matching and link discovery systems, and the management of provenance for Linked Data.

Foreword 6
Preface 8
Organization of the Book 9
Target Audience 10
Acknowledgments 12
Contents 13
About the Authors 16
1 Introduction 18
1.1 Semantic Web 18
1.2 Linked Data 22
1.3 Book Roadmap 24
2 Fundamentals 26
2.1 Linked Data 26
2.2 RDF 29
2.3 SPARQL 33
2.4 OWL 36
2.5 Reasoning 38
2.6 OWL 2 Profiles 42
2.7 Modern Big Data Storage and Processing Systems 43
2.7.1 NoSQL Databases 43
2.7.2 MapReduce/Hadoop 45
2.7.3 Spark 47
3 Centralized RDF Query Processing 50
3.1 RDF Statement Table 50
3.2 Index Permutations for RDF Triples 53
3.3 Property Tables 57
3.4 Vertical Partitioning 59
3.5 Graph-Based Storage 61
3.6 Binary Encoding for RDF Databases 65
4 Distributed RDF Query Processing 67
4.1 NoSQL-Based RDF Systems 67
4.2 Hadoop-Based RDF Systems 71
4.3 Spark-Based RDF Systems 77
4.4 Main Memory-Based Distributed Systems 79
4.5 Other Distributed RDF Systems 82
4.6 Federated RDF Query Processing 90
5 Processing of RDF Stream Data 100
5.1 RDF Streaming Data in A Nutshell 100
5.2 Data Representation of RDF Streams 103
5.3 RDF Streaming Query Model 105
5.3.1 Stream-to-Stream Operator 106
5.3.2 Stream-to-Relation Operator 106
5.3.3 Relation-to-Relation Operator 107
5.4 RDF Streaming Query Languages and Syntax 109
5.5 System Design and Implementation 111
5.5.1 Design 111
5.5.2 Implementation Aspects 113
5.5.2.1 Time Management 113
5.5.2.2 Scheduling and Handling Memory 115
5.5.3 Systems 116
5.5.3.1 Streaming SPARQL 117
5.5.3.2 C-SPARQL 118
5.5.3.3 EP-SPARQL 120
5.5.3.4 SPARQLstream 121
5.5.3.5 CQELS 121
6 Distributed Reasoning of RDF Data 124
6.1 The Process of RDF Reasoning 124
6.2 Peer-to-Peer RDF Reasoning Systems 127
6.3 NoSQL-Based RDF Reasoning Systems 131
6.4 Hadoop-Based RDF Reasoning Systems 132
6.5 Spark-Based RDF Reasoning Systems 135
6.6 Shared Memory RDF Reasoning Systems 137
6.7 Influence on Other Semantic Web Languages 139
7 Benchmarking RDF Query Engines and Instance Matching Systems 142
7.1 Benchmark Definition and Principles 142
7.1.1 Overview 142
7.1.2 Benchmark Development Methodology 144
7.1.3 Choke Points 145
7.2 Benchmarks for RDF Query Engines 147
7.2.1 Real Benchmarks 148
7.2.1.1 UniProt 148
7.2.1.2 YAGO (Yet Another Great Ontology) 149
7.2.1.3 Barton Library 149
7.2.2 Synthetic RDF Benchmarks 152
7.2.2.1 Lehigh University Benchmark (LUBM) 152
7.2.2.2 SP2Bench 154
7.2.2.3 Berlin SPARQL Benchmark (BSBM) 156
7.2.2.4 Semantic Publishing Benchmark (SPB) 161
7.2.3 Benchmark Generators 167
7.2.3.1 DBPedia SPARQL Benchmark (DBSB) 167
7.2.3.2 Waterloo SPARQL Diversity Test Suite 169
7.2.3.3 FEASIBLE 171
7.2.4 Dataset Structuredness 172
7.3 Benchmarks for Instance Matching Systems 174
7.3.1 Datasets 176
7.3.2 Variations 176
7.3.3 Reference Alignment 177
7.3.4 Key Performance Indicators 178
7.3.5 Real Benchmarks 178
7.3.5.1 A-R-S 2009 178
7.3.5.2 Data Interlinking (DI) 2010 180
7.3.5.3 Data Interlinking (DI) 2011 181
7.3.5.4 Overall Evaluation of Real Benchmarks 181
7.3.6 Synthetic Benchmarks for Instance Matching Systems 182
7.3.6.1 IIMB 2009 182
7.3.6.2 IIMB 2010 184
7.3.6.3 Person-Restaurants (PR) 2010 187
7.3.6.4 IIMB 2011 188
7.3.6.5 Sandbox 2012 188
7.3.6.6 IIMB 2012 189
7.3.6.7 RDFT 2013 189
7.3.6.8 ID-REC 2014 190
7.3.6.9 SPIMBench 2015 190
7.3.6.10 ONTOlogy Matching Benchmark with Many Instances (ONTOBI) 191
7.3.7 Overall Evaluation of Synthetic Benchmarks 192
7.4 Instance Matching Benchmark Generators for Linked Data 192
7.4.1 SWING 192
7.4.2 SPIMBENCH 193
7.4.3 LANCE 194
8 Provenance Management for Linked Data 195
8.1 An Overview of Provenance Models 195
8.2 Provenance Representations 197
8.3 Provenance Models 198
8.3.1 Relational Provenance 198
8.3.2 RDF Provenance 199
8.3.3 Update Provenance 202
8.4 Provenance in Data Management Systems 204
9 Conclusions and Outlook 210
9.1 Conclusions 210
9.2 Outlook 213
References 216