Integration of the Image and NL-text Analysis/Synthesis Systems

Applied Ontology

By ontology, we imply a conceptual “model of world”. Applied ontologies include concepts related both to ontology of tasks and ontology supporting knowledge representation in some practical application. Ontological engineering embraces: introducing classes of concepts and concept taxonomies, developing concept structures and situations, giving concept properties (attributes) and their ranges, working out inference procedures based on ontology and procedures of transforming concept relations into ontology.

If ontology is taken as an interface of integrated systems, then two problems arise. The first problem concerns the choice of ontological knowledge representation language such that this language would subsequently make it possible “to immerse” in it the required expansions, for example, fuzzy knowledge, cognitive procedures, etc. The second problem relates to the development of general applied ontology, capable, at the conceptual level, to integrate heterogeneous entrances for the synthesizers and outputs of the analyzers of the integral system. It is, of course, assumed that “inside” each sub-system there can be appropriate knowledge representation languages and bases of “internal” knowledge. For example, the Image Analyzer uses specific procedures or rules for extracting “non-derivative” (= primitive) objects; the Analyzer of NL-text uses specific methods for morphological, syntactical, and semantic NL-language processing.