Forecasting the transformation of professional competencies based on patent information and scientific publications

rapid technological development and the transition to a knowledge-based economy create a high degree of uncertainty regarding the future structure of employment and the required qualifications. Traditional labor market forecasting tools are oriented towards retrospective analysis and short-term perspectives, which does not allow for timely detection of technological shifts at the early stages of their emergence. The article presents an original approach to forecasting changes in professional competencies, built upon a Russian adaptation of a foreign method that uses text vector representations to establish links between occupational domains and patent activity. Unlike the original method, the proposed framework additionally takes into account scientific publication data, making it possible to detect emerging technological directions. The data sources include the Register of Professional Standards, the International Patent Classification, and the classifier of the Russian Institute of Scientific and Technical Information. A unified semantic space is formed using a word embedding model trained on a combined corpus of texts from professional standards, descriptions of patent classes, and scientific rubrics. The alignment of professional functions with technological classifiers, together with the analysis of patent and publication activity dynamics, enables the construction of two independent forecast tracks: a short-term track with a horizon of 3–7 years, reflecting changes in already established professions, and a long-term track with a horizon of 10–15 years, aimed at identifying competencies that may become in demand in the future. The proposed approach is theoretical in nature and does not replace existing forecasting methods but rather complements them by providing an additional objective channel of information about technological changes at early stages. Further research should be directed towards experimental testing of the proposed approach and its validation on a representative dataset in order to assess the adequacy of the forecast.

1. Glazyev, S. Yu. (1993), Theory of long-term technical and

2. economic development. VlDar, Moscow, Russia.

3. Perez, C. (2011), Technological revolutions and financial

4. capital: The dynamics of bubbles and golden ages. Translated from English, Delo, Moscow, Russia.

5. Pulman, A. L. and Puzyrevskaya, A. A. (2021), “Delphi

6. method of expert assessments”, Science Time, no. 5,

7. pp. 105–106.

8. Senthurvelautham, S. and Senanayake, N. (2023),

9. “A machine learning-based job forecasting and

10. trend analysis system to predict future job markets using historical data”, in Proceedings of the

11. th IEEE International Conference for Convergence

12. in Technology 2023 (I2CT 2023), Lonavla, India,

13. –9 April 2023. Piscataway: IEEE, 10126233. DOI:

14. 1109/I2CT57861.2023.10126233.

15. Bakhshi, H., Downing, J. M., Osborne, M. A. and Schneider, P. (2017), The future of skills: Employment in 2030.

16. Pearson, London, UK.

17. Gruetzemacher, R., Paradice, D. and Lee, K. B. (2020),

18. “Forecasting extreme labor displacement: A survey of

19. AI practitioners”, Technological Forecasting and Social

20. Change, vol. 161, pp. 120323.

21. Webb, M. (2019), “The Impact of Artificial Intelligence on the

22. Labor Market”, SSRN Electronic Journal. available at: https://

23. papers.ssrn.com/sol3/papers.cfm?abstract_id=3482150

24. (Accessed 7 May 2026) DOI: 10.2139/ssrn.3482150.

25. Ha, T., Lee, M., Yun, B. and Coh, B.-Y. (2022), “Job

26. forecasting based on the patent information: A word

27. embedding-based approach”, IEEE Access, vol. 10,

28. pp. 7223–7233.

29. Ha, T. and Moon, A. (2025), “Job demand estimation

30. using text embeddings of patent classification codes

31. and occupational data”, IEEE Access, vol. 13, pp. 1–14.

32. Rotolo, D., Hicks, D. and Martin, B. R. (2015), “What is

33. an emerging technology?”, Research Policy, no. 44 (10),

34. pp. 1827–1843. DOI:10.1016/j.respol.2015.06.006.