State fire service of EMERCOM of Russia (department of fire safety of transport of the Scientific research institute for advanced research and innovative technologies in the field of life safety, chief researcher)
Russian Federation
from 01.01.2023 until now
Voronezh, Russian Federation
employee from 01.01.2017 to 01.01.1925
UDC 004.62
This paper presents a comparative analysis of two architectures for decision support systems (DSS) aimed at adapting the content of educational programs to current labor market demands. The first architecture relies solely on the RuBERT language model to semantically match competencies from educational programs with requirements extracted from job postings. The second architecture extends this approach by integrating an ontological knowledge graph of skills and occupations based on the ESCO taxonomy. Experimental evaluation was conducted using real-world data from a Russian undergraduate program in applied informatics and a corpus of IT-related job vacancies. The results show that while the RuBERT-based architecture achieves high semantic matching accuracy, the ontology-enhanced system provides greater coverage of relevant skills (75% vs. 65%), more interpretable recommendations, and the ability to perform logical inference, such as identifying missing skill categories and aligning educational profiles with specific occupations. The incorporation of the Revealed Comparative Advantage (RCA) metric enables prioritization of curriculum updates based on skill demand intensity. The scientific novelty lies in the integration of pre-trained language models with a formalized skill ontology for curriculum design. The practical significance is demonstrated by the system’s ability to support curriculum developers, educational administrators, and accreditation bodies in continuously aligning educational content with dynamic labor market needs.
decision support system, educational programs, natural language processing, skills ontology, competence
1. Vasilyeva, E. A. A study of the discrepancy between graduates' competencies and labor market requirements // Economics of education. - 2023. — No. 3. — pp. 45-59.
2. Kozhevnikov, I. S. Comparative analysis of the effectiveness of natural language processing models for adapting educational programs to the requirements of the labor market / I. S. Kozhevnikov, V. F. Dubrovsky // Automation in industry. – 2025. – No. 4. – pp. 55-60. - FVPNAD PUBLISHING HOUSE.
3. Nikolaev A., Petrova N., Ivanov D. Extracting skills from Russian vacancies using domain-adaptive BERT // Materials of the AIST seminar 2023. - Materials of the CEUR seminar, volume 3450. - 2023.
4. ruWordNet: Thesaurus for the Russian synonymous structure based on WordNet. URL: https://ruwordnet.ru /
5. ESCO-European skills, competencies, qualifications and professions. URL: https://ec.europa.eu/esco/
6. O* NET Resource Center. URL: https://www.onetcenter.org/
7. Obolensky A. A., Shevchenko I. P. I'm going to say: "I'm going to go to the headquarters..................." I'm going to head to the headquarters. - 2023. — No. 5. - pp. 78-86.
8. Karakolis D., Kalamatianos J., Tsakalidis A. et al. Support for curriculum development through labor market analysis: the Quali-Chain platform / / Computers in Education Journal. — 2022. - Vol. 12, No. 4.
9. Nikolaev A. Adaptive skill recommendation system for HR specialists using knowledge graphs // Information technologies in education and management. — 2023.
10. DeepPavlov: An open source library for NLP tasks. URL:
11. Yu R., Das S., Gurajada S., Varshni K. R., Raghavan H., Lastra-Anadon S. H. A research framework for understanding the relationship between education and profession using NLP methods // Proceedings of the 1st NLP Seminar for Positive Impact (ACL). 2021. Pp. 100-106.
12. Fayzrakhmanov, R. A. Algorithm for extracting key concepts from educational programs of IT specialists using the hybrid contextual ranking method / R. A. Fayzrakhmanov, E. V. Dolgova, I. I. Sukhoi // Software Engineering. – 2025. – Vol. 16, No. 7. – pp. 334-346. - Doihttps://doi.org/10.17587/rare.16.334-346. - EDN RMARAZ.
13. Kozhevnikov, I. S. Comparative analysis of the effectiveness of natural language processing models for adapting educational programs to the requirements of the labor market / I. S. Kozhevnikov, V. F. Dubrovsky // Automation in industry. – 2025. – No. 4. – pp. 55-60. - FVPNAD PUBLISHING HOUSE.
14. Cheremukhin, A.D. Analysis of the structure of mathematical training of specialists in the field of public administration based on the study of curricula / A.D. Cheremukhin. Bachelor's Degree / A.D. Cheremukhin, N. N. Kolodkina // Samara Scientific Bulletin. – 2025. – Vol. 14, No. 2. - pp. 211-216. - DOIhttps://doi.org/10.55355/snv2025142312. - ELECTRONIC number PWPXFC.
15. Nadzinski G., Gerazov B., Zlatinov S., Kartalov T., Markowska-Dimitrovska M., Georski H., Chavdarov R., Kokolanski Z., Atanasov I., Horstman J., Sterle U., Gams M. The practice of teaching data science and machine learning with an emphasis on vocational education and training // Informatics in education. 2023. Volume 22, No. 4. pp. 671-690.
16. Stefanyak J., Xu M. A study of the systemic coverage of educational design models: a systematic review // TechTrends. 2020. Volume 64, No. 5. pp. 710-719.
17. NLP–based management of large repositories of multiple-choice test tasks / V. Albano, D. Firmani, L. Laura [et al.]/ / Journal of Learning Analytics. - 2023. - Volume 10, No. 3. - pp. 28-44. - DOIhttps://doi.org/10.18608/jla.2023.7897. - ED. QWVOJU.
18. Shaheen, Z. RuLegalNER: a new dataset for Russian legal named entities recognition / Z. Shaheen, D. I. Mouromtsev, I. Postny // Scientific and Technical Journal of Information Technologies, Mechanics and Optics. – 2023. – Vol. 23, No. 4. – P. 854-857. – DOIhttps://doi.org/10.17586/2226-1494-2023-23-4-854-857. – EDN CVZMBH.
19. Tomaszuk, D. PGO: Describing Property Graphs in RDF / D. Tomaszuk, R. Angles, H. Thakkar // IEEE Access. – 2020. – Vol. 8. – P. 118355-118369. – DOIhttps://doi.org/10.1109/access.2020.3002018. – EDN FSAHJT.
20. A scoping review of auto-generating transformation between software development artifacts / D. Siahaan, R. Fauzan, A. Widyadhana [et al.] // Frontiers in Computer Science. – 2024. – Vol. 5. – DOIhttps://doi.org/10.3389/fcomp.2023.1306064. – EDN AIYGRK.
21. Zhou, T. Semantic Relatedness Enhanced Graph Network for aspect category sentiment analysis / T. Zhou, K. M. Y. Law // Expert Systems with Applications. – 2022. – Vol. 195. – P. 116560. – DOIhttps://doi.org/10.1016/j.eswa.2022.116560. – EDN LFVJBO.
