Model of an Intelligent Agent Based on Formal Concept Analysis and DeepSeek for the Tasks of a Regional Spatial Development Centre

The article proposes a conceptual model of an intelligent agent that combines the method of Formal Concept Analysis (FCA) with the large language model DeepSeek. FCA formalises concepts and analyses data in the form of “object-attribute”. The article presents the theoretical foundations of FCA — formal contexts, concept lattices, Galois connections — and describes the capabilities of DeepSeek. The proposed hybrid model provides solutions to three classes of spatial development tasks: clustering of territorial objects, identification of implicative dependencies in spatial data, and decision support.  

1. Ganter B., Wille R. Formal Concept Analysis: Mathematical Foundations. Berlin: Springer, 1999, 284 p.

2. Wille R. Restructuring lattice theory: An approach based on hierarchies of concepts. Ordered Sets. Ed. I. Rival. Dordrecht: Springer, 1982, pp. 445–470.

3. Su F. Iskusstvennii intellekt bolshikh modelei kak draiver transformatsii kachestvennoi proizvoditelnosti v stroitelnoi otrasli: mekhanizmi, puti i prakticheskie issledovaniya na primere DeepSeek [Large model artificial intelligence as a driver of qualitative productivity transformation in the construction industry: mechanisms, paths and practical studies using DeepSeek]. Stroitelnie i dorozhnie mashini [Construction and Road Machinery], 2024, vol. 68, no. 3–4, pp. 40–52.

4. Andrews S. i et al. Formal Concept Analysis. Strategic Intelligence Management, 2013, pp. 209 233 (accessed: 12.05.2026).

5. Ignatov D.I. Introduction to Formal Concept Analysis and Its Applications in Infor-mation Retrieval and Related Fields. 2017. Available at:https://arxiv.org/abs/1703.02819 (accessed: 12.05.2026).

6. Guigues J.L., Duquenne V. Familles minimales d'implications informatives résultant d'un tableau de données binaires. Mathématiques et Sciences humaines, 1986, no. 95, pp. 5–18.

7. Denniston J.T., Melton A., Rodabaugh S.E. Formal Contexts, Formal Concept Analysis, and Gal ois Connections, 2013. Available at: https://arxiv.org/abs/1309.5134 (accessed: 12.05.2026).

8. Timashev S. Artificial Intelligence in Risk-Based Design and Operation of Infrastructures: Dangers and Opportunities. International Journal for Computational Civil and Structural Engineering, 2025, vol. 20, no. 11. Available at: https://cyberleninka.ru/article/n/artificial-intelligence-in-risk-based-design-and-operation-of-infrastructures-dangers-and-opportunities

9. DeepSeek LLM Technical Documentation. DeepSeek AI, 2024–2025. Available at: https://github.com/deepseek-ai (accessed: 02.06.2026).

10. Rospatent odobril registratsiyu tovarnogo znaka populyarnogo v Rossii kitaiskogo II-startapa [Rospatent approved the registration of the trademark of the Chinese AI start-up popular in Russia]. Informatsionnoe agentstvo SakhaNews / TASS [SakhaNews Information Agency / TASS], 2026, 4 March. Available at: https://sakhanews.ru (accessed: 25.04.2026).

11. Kitaiskaya kompaniya DeepSeek registriruet svoi tovarnie znaki v Rossii [Chinese company DeepSeek registers its trademarks in Russia]. Agentstvo gorodskikh novostei “Moskva” [Moscow City News Agency], 2025,

12. February. Available at: https://mskagency.ru (accessed: 12.05.2026).

13. BRIDGE: Benchmarking Large Language Models for Understanding Real-world Clinical Practice Text, 2025. Available at: https://arxiv.org/abs/2504.19467 (accessed: 12.05.2026).

14. Kuleshova Yu.I. Neironnaya set DeepSeek i yee primenenie [The DeepSeek neural network and its applications]. Lomonosov-2025 : materiali Mezhdunarodnoi nauchnoi konferentsii studentov, aspir antov i molodikh uchenikh [Lomonosov-2025 : materials of the International Scientific Conference of Students, Postgraduates and Young Researchers]. Moscow: MGU, 2025, pp. 120–125.

15. Klimova S.G., Mikheenkova M.A., Finn V.K. DSM-metod v kachestvennom sotsio-logicheskom issledovanii: osnovnie printsipi i opit ispolzovaniya [DSM-method in qualitative sociological research: basic principles and experience of use]. Sotsiologicheskii zhurnal [Sociological Journal], 2016, vol. 22, no. 2, pp. 8–30. DOI: 10.19181/socjour.2016.22.2.4255.

16. Finn V.K. DSM-metod kak sredstvo analiza kauzalnikh zavisimostei v intellektualnikh sistemakh [DSM-method as a tool for analysing causal dependencies in intelligent systems]. Nauchno tekhnicheskaya informatsiya. Seriya 2. Informatsionnie protsessi i sistemi [Scientific and Technical Information. Series 2. Information Processes and Systems], 2000, no. 11.

17. Finn V.K., Mikheenkova M.A. O situatsionnom rasshirenii DSM-metoda avtomaticheskogo porozhdeniya gipotez [On the situational extension of the DSM-method of automatic hypothesis gen eration]. Nauchno-tekhnicheskaya informatsiya. Seriya 2. Informatsionnie protsessi i sistemi [Scien tific and Technical Information. Series 2. Information Processes and Systems], 2000, no. 11, pp. 20–28.

18. Finn V.K. Evristika obnaruzheniya empiricheskikh zakonomernostei i printsipi intellektualnogo analiza dannikh [Heuristics for discovering empirical patterns and principles of data mining]. Iskusstvennii intellekt i prinyatie reshenii [Artificial Intelligence and Decision Making], 2018, no. 3, pp. 3–19. DOI: 10.14357/20718594180311.

19. Kuznetsov S.O., Obiedkov S.A. Comparing Performance of Algorithms for Generat-ing Concept Lattices. Journal of Experimental & Theoretical Artificial Intelligence, 2002, vol. 14, no. 2–3, pp. 189–216.

20. DeepSeek License Agreement. Version 1.0. GitHub: deepseek-ai/DeepSeek-V2. Available at: https://github.com/deepseek-ai/DeepSeek-V2/blob/main/LICENSE-MODEL (accessed: 25.04.2026).