Welcome to the official page of the Autonomous Non-Profit Organization «Center for Scientific Research in Career Guidance and Occupational Psychology» (ANO «CRCG»).
We function as an independent research hub dedicated to transitioning the field of career guidance and vocational counseling from a domain of speculative humanities into a rigorous, data-driven science.
What We Do
We conduct interdisciplinary research at the intersection of neuroscience, cognitive science, psychology, machine learning (ML), and big data analytics.
Our core research framework focuses on three fundamental areas:
- Psychology of Self-Determination
We investigate the cognitive and behavioral mechanisms of personal, professional, familial, political, and religious self-determination. Our objective is to map these underlying processes to develop actionable guidance tracks for both adolescents and adults. Methodologically, we bridge the gap between neurobiology (specifically, Karl Friston’s Free Energy Principle and predictive coding frameworks) and higher mental activity. This is achieved through formal mathematical modeling as well as applied intervention strategies—ranging from structured self-determination counseling to specialized eduLARP (Educational Live Action Role-Playing) environments where choices are enacted through direct action. - Taxonomy of the World of Work and Occupations
Our team has developed a comprehensive Meaning-Based Classification of Occupations. This structural framework comprises dynamic taxonomies of industries, working conditions, subjects of labor, and professionally important qualities (PIQs), alongside the evolutionary laws governing professional tools. - Digital Labor Market Modeling
We design dynamic computational models where job vacancies and resumes are analyzed not as static snapshots, but as continuous flows on a timeline. This model is continuously enriched via automated data pipelines pulling from diverse external sources, including educational platforms, state statistics, tax registries, geographic information systems (GIS), and corporate directories.
Our Principles: Open Science & Academic Freedom
We operate under an organizational model designed to ensure maximum intellectual independence and minimize external bureaucratic pressures:
- Long-Term Funding Architecture. The Center has opted out of the traditional short-term academic grant system. Our fundamental research operations are secured via multi-year, long-term funding provided directly by our founder (Paramult Company). This structural independence frees our investigators from chasing immediate funding cycles, allowing them to maintain rigorous scientific objectivity.
- Focus on Substance Over Metrics. As an organization, the Center does not participate in the race for institutional citation quotas or artificial publication inflation. Instead, we prioritize the actual reproducibility and real-world impact of our findings. At the same time, we fully encourage and support our team members in their independent academic pursuits and peer-reviewed publications.
- The Open Science Commitment. Our ultimate goal is the generation and dissemination of actionable knowledge. We actively publish the methodologies, source codes, and results of our research in the open domain.
Core Values of Our Research Ethics
- Intellectual Honesty. A null result or a failed hypothesis is a valuable scientific outcome.
- Reproducibility. Research workflows must be structured so that any colleague can replicate the results.
- Healthy Skepticism. We maintain a culture of rigorous, constructive critique focused strictly on ideas, never on individuals.
- Methodological Parsimony. We adhere to Occam’s razor, avoiding unnecessary complexity where simpler models suffice.
- Collaboration. We approach modern science as a highly collaborative, interdisciplinary team sport.
- Cognitive Bias Management. We actively implement protocols to mitigate confirmation bias and systemic cognitive distortions in data analysis.
- Social Responsibility. We evaluate our research through the lens of its ethical implications and long-term societal impact.
Applied Impact & Partnerships
To ensure that our theoretical breakthroughs result in practical tools, the Center collaborates with leading educational initiatives and edtech platforms.
Our mathematical models and taxonomies serve as the scientific foundation for:
- The Navigatum career-guidance platform—a widespread applied educational ecosystem currently operating across a network of over 3,000 experimental hubs in Eastern Europe and Central Asia.
- The Iz-I personal self-determination digital ecosystem, designed to provide scalable counseling tools for wider audiences.
Funding & Transparency
Our financial model is divided into two distinct, transparent streams:
- Fundamental Theoretical Research: Secured via institutional multi-year funding, enabling deep, uninterrupted scientific exploration of cognitive models.
- Applied Tool Development: Co-funded and executed in partnership with commercial and non-profit educational entities that deploy our research in real-world environments.
Selected Research Projects
Recent preprints and publications:
📄 Cartography of the Mental Topography of Free-Energy Minimization: A Computational Approach Based on Narrative Trajectory Analysis Using Large Language Models (LLMs)
Abstract: How does the brain protect its beliefs from contradictory facts? Neuroscience answers: through free-energy minimization (FEP, K. Friston), but it does not know the content of these defenses. Linguistics is able to analyze text topics but does not see the dynamics of cognitive tricks. We propose an interdisciplinary method that extracts from text corpora not static topics but dynamic transition graphs: from initial prior beliefs through a point of cognitive friction to final smoothing. Using LLMs in deep reasoning mode, we filter markers of cognitive effort, identify defense operators, and classify them by energy cost (external vs. internal attribution). A rigorous falsifiable hypothesis is advanced concerning a power-law rank distribution of operators, in which low-cost operators (externalization, escapism) monotonically dominate high-cost operators (reflection, restructuring of priors). The method includes a stratified sample across three genres, two-pass semantic compression guided by a taxonomy (Kahneman, A. Freud), and a second-derivative algorithm for identifying the defense core. The project translates qualitative hypotheses into measurable variables and creates the first empirical atlas of collective cognitive algorithms.
Keywords: free-energy principle, predictive coding, cognitive defenses, large language models, narrative analysis, frequency hierarchy, falsifiability
🔗 doi.org (Available in English)
📄 E-CDP: Extended Controlled Deformation Protocol for Assessing the Structural Significance of Semantic Domains in Text Corpora Using Large Language Models
Abstract: This note presents a formal description of the E-CDP (Extended Controlled Deformation Protocol), an experimental protocol designed to address challenges in the digital humanities. The protocol employs large language models (LLMs) as a tool for the quantitative analysis of text corpora, rather than as an object of study in itself. The core idea is as follows: using a target corpus, several versions of a training dataset are created — one unchanged (control) and three deformed versions for each semantic domain under investigation (semantic removal, introduction of logical contradiction, and stylistic paraphrase as a control). A base LLM is independently fine-tuned on each version using identical hyperparameters, yielding a family of equivalent models. The changes induced by these deformations are assessed using a system of binary questions with expert-predefined answer keys. The quantitative metric — the Structural Significance Coefficient (SSC) — captures how critical a domain is for preserving the integrity of the discursive worldview embedded in the corpus. This note sets out the methodological architecture of the protocol; all implementation details, examples, program code, and current experimental results are published sequentially in the online appendix.
Keywords: digital humanities, large language models, fine-tuning, latent space, controlled deformation, isomorphic deformation, structural significance coefficient, semantic domain, E-CDP protocol, computer experiment
🔗 doi.org (Available in English)
📄 Empirical substantiation of the thematic core of personal self-determination: results of protocol implementation on a Russian-language corpus
Abstract: The article presents the results of the empirical implementation of a protocol for extracting a thematic core intended for the pedagogical support of adolescents’ personal self-determination (PSD). Using a corpus of 100,223 annotations of Russian-language literary works, the study employed the large language model Gigachat3.1-10b and a three-stage post-processing procedure utilizing the DeepSeek model to obtain a structural-register classification of domains of human experience. The resulting framework comprises 36 domains and 243 themes. The article describes deviations from the initial protocol, the procedures for data normalization and filtering, and the rationale for opting for LLM-based semantic analysis over automated clustering. The final classification is compared with the expert list «Facets of Personality,» revealing an 83–84% overlap in theme coverage. Supporting materials (the classification scheme, discussion questions, code, and prompts) are provided in the appendices to the article.
Keywords: personal self-determination, thematic analysis, large language models, corpus linguistics, semantic classification, pedagogical support
🔗 doi.org (Available in Russian)
Leadership & Research Team
Director of Research
Anton U. Smirnov (Academic Profile)
Anton focuses on bridging fundamental cognitive frameworks with applied educational platforms. As a chief methodologist, he specializes in instructional design, developing advanced training programs, eduLARP frameworks, corporate educational simulations, and VR-based behavioral applications. All developed tools are strictly anchored in verified empirical data and structured educational modeling.
🌐 Academic Profiles: ORCID | Google Scholar
The Team
Our interdisciplinary core unites machine learning engineers, Big Data specialists, methodologists, cognitive scientists, and immersive environment developers. We build breakthrough tools at the intersection of our respective competencies.
Academic & Partnership Opportunities
The Center regularly invites postdoctoral researchers, PhD candidates, and independent scholars who share our core values to conduct collaborative or autonomous research.
- We minimize administrative overhead, providing our partners with the computational infrastructure, data pipelines, and funding required to execute high-level scientific projects.
- We offer internal institutional grants to fully fund promising research initiatives.
- Our partners enjoy absolute freedom from rigid institutional KPIs, direct access to proprietary datasets, and the unique opportunity to see their theoretical models implemented in large-scale applied ecosystems.
📩 Contact for academic inquiries: welcome@crcg.ru
Legal Note: Due to variations in regional registration translation standards, the Center is also occasionally indexed in historical archives under the alternative legal name form: ANO «Center for Scientific Research in the Field of Career Guidance and Labor Psychology».