Generative ontology-controlled platform of micro-dictionaries: from the preparation of text corpses to educational “capsules of knowledge”

The paper presents the outcomes of the development the generative, ontology-driven Microdictionary platform as part of the POLYHEDRON ecosystem. The platform enables an end-to-end workflow “corpus → microdictionary → educational course”, covering: consolidation, cleaning, and normalization of heterogeneous textual sources; creation of structured dictionary entries as compact “knowledge capsules”, and analytics of their educational use. Two interconnected subsystems are described: a text-preparation pipeline for microdictionary corpora and the transdisciplinary “MediaCapsules” cluster; the generative ontology-driven Microdictionary platform that integrates ontological modeling, corpus methods, GenAI-assisted drafting, and didactic tools. Microdictionaries are conceptualized as compact yet structurally complete resources: each entry includes a definition with references, corpus-based examples, an author’s didactic commentary, multimedia elements, and ontology links. The platform supports L1/L2 classification, quality control for AI-generated drafts, and both static and dynamic analytics (coverage, source linkage, usage indicators in educational activities). Pilot educational scenarios are reported (media literacy and information security contexts, youth slang, and programming terminology), together with evaluation signals for ontology classification quality, usefulness of generative drafts, and completeness of knowledge capsules. The paper outlines future directions, including domain expansion of ontologies, integrity-preserving integration with learning assessment, scaling of the “MediaCapsules” cluster, and international collaboration towards multilingual microdictionaries.