Mediating interdiscursive professional communication requires assessing the cognitive proximity of domains. Such an assessment may be subjective, labor-intensive and expert-oriented, or fail to reflect the actual state of domains’ development when based on predefined ontologies. The study aims to develop software for the automated assessment of domains’ cognitive proximity based on the analysis of up-to-date terminological corpora. The proposed approach is implemented through an interactive ontological graph service that combines statistical methods for term extraction and processing with SBERT semantic embeddings to construct vector representations of terminologies and calculate the degree of semantic proximity between domains. A resulting ontological graph provides interactive visualization of the semantic relationships between subject areas. The software supports processing of Russian- and English-language corpora and automates the entire analytical workflow, from collecting texts from Wikipedia and arXiv to visually presenting the results, which significantly reduces the time required to build ontological models compared to semi-automated methods. A comparative analysis of the algorithm variants showed that linguistic and statistical term processing improves the differentiation between domains, while cross-domain IDF weighting increased the range of similarity scores by approximately 70–73% compared with the baseline. The practical significance of the research consists in the development of a reproducible prototype for analyzing interdisciplinary connections, which can be used in terminology studies, ontology development, scientific knowledge analysis, and the preparation of teaching materials.