Instrumental methods and models for constructing a benchmark profile of personnel competencies in organisations
Vitaliy A. Kudrevatykh - Postgraduate Student. Perm State University
Abstract
Modern systems for assessing personnel competencies are highly labour-intensive and subjective. The paper aims to develop a methodology and a system for the automated formation of benchmark competency profiles based on the analysis of job vacancy data from the Russian labour market utilising natural language processing (NLP) methods and large language models (LLM). The study relies on an analysis of 78,882 job vacancies across 269 professional roles from the “hh.ru” platform. The research applies the locally deployed LLM model “gemma3:12b-it-qat” to extract competencies, “ai-forever/sbert_large_nlu_ru” embeddings for deduplication, and frequency analysis to identify relevant requirements. The author has developed a system for creating structured competency profiles. The combined approach (LLM, word embedding, and frequency ranking) identifies significant professional requirements in terms of specialties and qualification levels. The example of the profession “Data Scientist” demonstrates the differentiation of skills: from basic instrumental knowledge for junior positions to system design competencies for senior specialists. The profiles enable real- time tracking of trends in the development of required competencies. The results confirm the effectiveness of big data and LLM methods for the objective assessment of labour market requirements. The profiles are useful for performance appraisal systems, educational programme development, and HR analytics. The study may be of interest to specialists in the field of human resource management and to educational organisations. A promising direction is the integration of additional data sources for assessing the actual level of personnel competencies.
Keywords: competency assessment; large language models; LLM; NLP; competency profiles; labour market; big data; data mining.
For citation: Kudrevatykh V. A. Instrumental methods and models for constructing a benchmark profile of personnel competencies in organisations. Digital Models and Solutions. 2025. Vol. 4, no. 4, pp. 32–48. DOI: 10.29141/2949-477X-2025-4-4-3. EDN: OYHPMW.
