Semantic alignment of ontologies meaningful categories with the generalization of descriptive structures

Abstract

The presented work addresses the issue of semantic alignment of ontology components with a generalized structured corpus. The field of research refers to the sphere of determining the features of trust in artificial intelligence. An alignment method is proposed at the level of semantic components of the general alignment system. The method is a component of a broader alignment system and compares entities at the level of meaningful correspondence. Moreover, only the alignment entities’ descriptive content is considered within the proposed technique. Descriptive contents can be represented by variously named id and semantic relations. The method defines a fundamental ontol- ogy and a specific alignment structure. Semantic correspondence in the form of information scope is formed from the alignment structure. In this way, an entity is formed on the side of the alignment structure, which would correspond in the best meaningful way to the entity from the ontology in terms of meaningful descriptiveness. Meaningful descriptiveness is the filling of information scope. Information scopes are formed as a final form of generalization and can consist of entities, a set of entities, and their partial union. In turn, entities are a generalization of properties that are located at a lower level of the hierarchy and, in turn, are a combination of descriptors. Descriptors are a fundamental element of generalization that represent principal content. Descriptors can define atomic content within a knowledge base and represent only a particular aspect of the content. Thus, the element of meaningfulness is not self-sufficient and can manifest as separate meaningfulness in the form of a property, as a minimal representation of the meaningfulness of an alignment. Descriptors can also supplement the content at the level of information frameworks, entities, and properties. The essence of the alignment in the form of information scope cannot be represented as a descriptor or their combination. It happens because the descriptive descriptor does not represent the content in the completed form of the correspondence unit. The minimum structure of representation of information scope is in the form of properties. This form of organization of establishing the correspondence of the semantic level of alignment allows you to structure and formalize the information content for areas with a complex form of semantic mapping. The hierarchical representation of the generalization not only allows simplifying the formalization of semantic alignment but also enables the formation of information entities with the possibility of discretization of content at the level of descriptors. In turn, descriptors can expand meaningfulness at an arbitrary level of the generalization hierarchy. This provides quantization of informational content and flexibility of the alignment system with discretization at the level of descriptors. The proposed method is used to formalize the semantic alignment of ontology entities and areas of structured representation of information.

Prombles in programming 2022; 3-4: 355-363