Application of constraint propagation techniques to speed up processing of queries to ontologies

In the paper, we propose to regard a problem of processing of SPARQL queries to the ontology as a constraint satisfaction problem. The formal apparatus for formalization of constraint satisfaction problems by means of specialized matrix-like structures is briefly given. The use of this apparatus allows one to describe, store and process non-numeric constraints of the subject domain more efficiently compared with the table representation. To speed up processing of queries to big ontologies it is proposed to apply the non-numeric constraint propagation technique previously developed by the authors. Compared with the traditional approach to processing of the SPARQL queries, which is based on dynamic programming, the proposed technique allows one to accelerate their performance by "compact" representation of ontology, as well as by using the original rules of the search space reduction. Application of the mathematical apparatus for reducing the dimension of the search space at processing a query to the ontology of multidisciplinary knowledge is exemplified.