We offered, in a previous paper, an ontology-based approach to recognize
constraints in free-form service requests and provide services for users. Our
system handles a service request by finding, from among many ontologies, the
domain ontology tha
t best matches the request and then uses the matched
ontology to generate the service request constraints. Although our system is
powerful in recognizing constraints and therefore servicing requests, the
recognition process is a bottleneck due to the number of the tested ontologies
and the amount of computations involved. This paper provides a novel
approach to speed up the recognition process by (1) using ontology indexing
and (2) excluding inapplicable regular expressions early in the process and thus
reducing the number of applied regular expressions. Our experiments show
that our techniques are effective in significantly reducing the amount of
computations and therefore speeding up the recognition process.
