Automatic generation of repair suggestions for overall I&C architecture represented with an ontology

We present an approach for suggesting possible fixes to an overall I&C nuclear architecture during its design phase. Despite the I&C architecture, in our case, being represented with an ontology, we do not aim to change the properties of an ontology per se. Instead, we focus on the subset of ABox triples that do not contain terminological elements in either subject, predicate, or object parts. Such a subset we call the design artifacts. When the ontology is filled with the design artifacts, the analyst runs the check of non-functional requirements using SPARQL queries. The requirements associated with the queries that returned results do not hold. The goal of the current work is to provide support for the next stage when the analyst has to change the design artifacts so that the queries no longer return results. Our method is based on representing the results of queries as graphs, intersecting them, and finding the minimal changes that prevent the results from being mapped on their (or other) queries.