Federated Data Management and Query Optimization for Linked Open Data

Linked Open Data provides data on the web in a machine readable way with typed links between related entities. Means of accessing Linked Open Data include crawling, searching, and querying. Search in Linked Open Data allows for more than just keyword-based, document-oriented data retrieval. Only complex queries across different data source can leverage the full potential of Linked Open Data. In this sense Linked Open Data is more similar to distributed/federated databases, but with less cooperation between the data sources, which are maintained independently and may update their data without notice. Since Linked Open Data is based on standards like the RDF format and the SPARQL query language, it is possible to implement a federation infrastructure without the need for specific data wrappers. However, some design issues of the current SPARQL standard limit the efficiency and applicability of query execution strategies. In this chapter we consider some details and implications of these limitations and presents an improved query optimization approach based on dynamic programming.