Abstract:
Approximate Reasoning involves compromising the soundness or the completeness of reasoning
over description logic ontologies in favor of significant speed up in the run time. The efficiency
of approximate Reasoning can further be improved by making the entailment rules amenable
for parallel processing i.e concurrent reasoning. Hence we focused on the rule based reasoning
where a set of entailment (also called as completion) rules are applied on the axioms in the
ontology to derive inferences in an approximate-concurrent manner.
Previous semester we carried out experiments to classify different ontologies of increasing sizes using different reasoners that support approximate reasoning, standard reasoning, and concurrent
reasoning to distinguish among them based on their individual performance. Our experiments
showed that concurrent classification and approximation deduction hugely impact the classification time with increasing size of ontologies. We aim to improve the efficiency of approximate
reasoning by extending the entailment rules amenable for parallel processing.
