Optimized Data Access for Efficient Execution of Semantic Services

Executing Semantic Services requires, in contrast totraditional SOAP-based Web Services, frequent read and writeaccesses to graph-based semantic data stores – for instance, forthe evaluation of preconditions or the materialization of serviceeffects. Therefore, the overall performance of semantic serviceexecution, in particular for composite services, is strongly affectedby the efficiency of these reads and writes. In this paper wepresent two data access optimization techniques for semanticdata stores: Prepared Queries and Frame Caching. The formerreduces the costs for repeated query evaluation, e.g., in loops.The latter provides rapid access to frequently read triples orsubgraphs based on materialized views using a Frame-based datastructure. The described techniques have been implemented andevaluated on the basis of OSIRIS NEXT, our open infrastructurefor Semantic Service support.