Our daily use of Internet and related technologies generates continuously large amount of heteregenous data flows. Several RDF Stream Processing (RSP) systems have been proposed. Existing RSP systems benefit from the advantages of semantic web technologies and traditional data flow management systems. C-SPARQL, CQELS, SPARQL stream , EP-SPARQL, and Sparkwave extend the semantic query language SPARQL and are examples of those systems. Considering that the storage and processing of all these streams become expensive, we propose a solution to reduce the load while keeping data semantics, and optimizing treatments. In this paper, we propose to extend C-SPARQL for continuously generating samples on RDF graphs. We add three sampling operators (UNIFORM, RESERVOIR and CHAIN) to the C-SPARQL query syntax. These operators have been implemented into Esper, the C-SPARQL's data flow management module. The experiments show the performance of our extension in terms of execution time and preserving data semantics.

No abstract

No abstract

No abstract