Learning from the History of Distributed Query Processing - A Heretic View on Linked Data Management

Published on 2012-12-032367 Views

Francis Gropengießer

The vision of the Semantic Web has triggered the development of various new applications and opened up new directions in research. Recently, much effort has been put into the development of technique

Third International Workshop on Consuming Linked Data

Related categories

Consumption

Presentation

LEARNING FROM THE HISTORY OF DISTRIBUTED QUERY PROCESSING00:00

Motivation00:12

Agenda01:28

DISTRIBUTED AND FEDERATED DATABASES01:53

Systems02:01

Techniques02:52

Still Unsolved Problems (1)03:09

Still Unsolved Problems (2)04:06

THESES04:12

1. The Volume of Linked Data Is Too Big for Centralized Management04:23

2. Materialization in Centralized Repositories Violates Data Authority05:50

3. Scalability Can Be Achieved Only by Distributed Query Processing06:55

4. Linked Data Processing Is only about SPARQL Processing07:41

5. The Problem of Semantic Heterogeneity Can Be Solved by Using Ontologies08:22

6. HTTP URIs Can Be Used to Identify Relevant Endpoints09:14

7. The Freshness of Data Is Guaranteed only with Distributed Query Processing09:59

8. Open Data Is Accessible as Linked Data10:53

9. Centralized Linked Data Is not Linked Data anymore11:34

RESEARCH AGENDA12:08

Linked Data as a Service12:25

1. Linked Data Processing beyond SPARQL Processing13:54

2. Exploit Newly Available Infrastructure/Platform as a Service14:26

3. Address the Opportunities of Modern Hardware Architectures for Query Processing15:06

4. Realistic Benchmarking and Metrics15:42

5. Simplify Publishing and Exploit Crowdsourcing16:27

CONCLUSION16:54

Conclusion 116:56

CLOSING WORDS17:44