Page 32 - 2019

unused nodes in the ontology The results have shown overload of 5 in SPARQL query
comparison using graph data model and significant increase in the performance of
SPARQL query execution because of reduced search space without compromising the
knowledgebase marking the unused nodes in ontologies to avoid traversal of these nodes
based on semantic measures. Then converting the SPARQL query into equivalent graph
and use the semantic similarity measures to compare SPARQL query under execution
with already executed SPARQL query in the system. Scope of search space can be
reduced in ontology by marking unused nodes in the ontology. The results have shown
overload of 5% in SPARQL query comparison using graph data model and significant
increase in the performance of SPARQL query execution because of reduced search
space without compromising the knowledge base. Our approach in SPARQL query
optimization uses semantic similarity measure for comparing SPARQL query. Clustering
the nodes within ontology based on semantic measures threshold set for SPARQL query
and semantically measured quantitative output. Mark the nodes which are used to
execute the query, its sub-class and all super class[1,2, 3]and stored in reference
repository for further reference. When second time the same or similar SPARQL query
request is received on the ontology, the query engine will refer to the specific markers
which are nothing but unused nodes within the ontology and skip these nodes. This
approach has been tested on manually pruned ontology and 30% enhancement in
executing the query is observed. We have proposed an integrated framework on dotnet
RDF which uses semantic similarity toolboxes before the query is executed and
respectively route it to specific cluster in the ontology.
Keywords: Semantic Web, OWL, SPARQL, Ontologies, RDF, Federation of nodes, scope of
search, Performance enhancement

13