Federated SPARQL basic graph pattern optimization using links over the web of linked data

The fast growth of the web of linked data raises new challenges for distributed query processing. Different from traditional federated databases, linked data sources cannot cooperate with each other. Hence, sophisticated optimization techniques are necessary for efficient query processing. In this paper, we formalize the problem of Basic Graph Pattern (BGP) optimization for federated SPARQL queries over the Web of Linked Data. We define and analyze the characteristics of source selection for links based static BGP optimization. The classes of bound subject and object associated with bound predicates of triple patterns are first used to select the set of relevant sources. Then links between linked data are used to prune the relevant sources of triple patterns. With the FedBench benchmark, we evaluate the performance of our approach of source selection for FedBench queries. The results of the evaluation show the feasibility of our approach.