Federated Query Processing Techniques On Linked Data

Since Berners-Lee gave a talk in the TED Conference 2009, Linked data has been rapidly developing. By September 2011 the Linked Open Data cloud had grown to 31 billion RDF triples, interlinked by around 504 million RDF links. How to effectively query the huge amount of linked data has been a hot research subject.The thesis designs the federated query processing structure on Linked Data, which is characterized by the creation of the statistical information. It presents the data source selection algorithm. Data source mapping generates corresponding sub-queries to different data sources. Inverted index is used locally to manage the relevant statistics. It is implemented by sending ASK queries to SPARQL endpoint using VOID described. Based on the SPARQL basic graph pattern, the heuristic query optimization uses FILTER equivalent rules to join order optimization. The cost-based optimization of linked data defines cost function and the federated cardinality estimation. Query plan enumeration is chosen the dynamic programming LDPccp algorithm ,which is the traditional relational database optimization strategies.FR is implemented based on the Sesame SAIL layer. Experiments which are used the state-of-the-art benchmark FedBench show that the federated query processing architecture is reasonable and the optimization techniques improve query processing performance.