METHODS OF SUPPORT, MANAGEMENT AND INTEGRATION OF HIGHLY DIVERSIFIED INFORMATION

Abstract

The paper introduces an ontology-driven methodology for semantic integration of heterogeneous educational resources into a unified knowledge environment built on the Resource Description Framework (RDF). Printed textbooks, slide decks, video lectures, laboratory manuals and other learning artefacts are decomposed into elementary “subject–predicate–object” triples. This representation enables the platform to (i) identify duplicate or semantically equivalent notions (e.g., “two-dimensional array” vs. “matrix”), (ii) detect terminology conflicts across independently authored materials, and (iii) reconstruct implicit prerequisite chains between concepts.

To quantify annotation, we define three formal indicators. V_n measures coverage, i.e. the ratio of unique informative fragments captured by triples to the potential number of knowledge units in the source. V_m reflects term alignment by comparing the set of classes and properties used in an annotation with those already present in the domain ontology. V_b estimates logical consistency as the share of statements that do not contradict previously validated facts. These coefficients are calculated automatically after every ingestion or revision step and serve as immediate feedback for instructional designers.

The methodology is validated through a “computer-to-computer” role-playing experiment. Two autonomous agents interact within the domain “Introduction to Programming”: the Generator assembles learning paths and produces candidate tests, while the Editor periodically adds new fragments or deliberately introduces terminology variations. A few cycles demonstrate how Vn steadily rises from 0.80 to 0.97 as the corpus expands, whereas temporary drops in Vm and V_b reliably signal emerging conflicts that are later resolved via ontology enrichment. The experiment confirms that the triple-based representation supports both incremental evolution of the knowledge graph and automated synthesis of personalized assessments without manual curation of item banks.

The proposed approach extends beyond static content organization: it lays the groundwork for adaptive e-learning systems capable of self-diagnosing knowledge gaps, recommending targeted resources, and continuously updating course structures in response to curricular changes.