N.V. Blokhin1, S.V. Makrushin2

1−2 Financial University under the Government of the Russian Federation (Moscow, Russia)

1−2 Center for Intelligent Analytical and Robotic Systems (Balashikha, Russia)

1nvblokhin@fa.ru, 2svmakrushin@fa.ru

A proper representation of the subject's industry sector is a key factor for solving a wide range of applied economic problems. At this point, All-Russian Classification of Economic Activities (OKVED) is used as the main industry sector description tool in Russia. OKVED is a hierarchical classifier that primarily uses the production process to determine the type of economic activity. At the same time, an analysis of the relationships between companies can give a more complete picture of the functioning of a particular industry or the positioning of industries relative to each other. It is proposed to develop a model for constructing vector representations of the graph classifier of economic activities, taking into account the relationship between companies which are representatives of industries. In this paper, the author's approach to the construction of a graph of economic activity codes, taking into account information about the relationships between companies, is proposed. In order to obtain embeddings, the link prediction problem for the relationships obtained on the basis of information about the relationships between companies is solved. To test the quality of the received embeddings of OKVED codes, an auxiliary problem of binary classification of the presence of profit (loss) of a company based on the growth rates of financial indicators has been solved. The obtained results demonstrate that the received embeddings of OKVED codes contain information that is useful solve forecasting problems more efficiently.

Blokhin N.V., Makrushin S.V. Construction of a vector representation of economic activities using graph neural networks. Information-measuring and Control Systems. 2023. V. 21. № 5. P. 7−15. DOI: https://doi.org/10.18127/j20700814-202305-02 (in Russian)

1. Prikaz Rosstata ot 31.12.2014 N 742 (red. ot 04.02.2016) "O Metodicheskikh ukazaniyakh po opredeleniyu osnovnogo vida ekonomicheskoi deyatelnosti khozyaistvuyushchikh sub'ektov na osnove Obshcherossiiskogo klassifikatora vidov ekonomicheskoi deyatelnosti (OKVED2) dlya formirovaniya svodnoi ofitsialnoi statisticheskoi informatsii". (in Russian)
2. Federalnyi zakon ot 08.12.2003 N 165-FZ (red. ot 08.12.2020) "O spetsialnykh zashchitnykh, antidempingovykh i kompensatsionnykh merakh pri importe tovarov". (in Russian)
3. Uryadnikova Marina Valerevna Osobennosti vliyaniya otraslevykh rynkov na tsenoobrazovanie. Vestnik ChGU. 2010. № 1. URL: https://cyberleninka.ru/article/n/osobennosti-vliyaniya-otraslevyh-rynkov-na-tsenoobrazovanie (data obrashcheniya: 12.06.2023). (in Russian)
4. "OK 029-2014 (KDES Red. 2). Obshcherossiiskii klassifikator vidov ekonomicheskoi deyatelnosti" (utv. Prikazom Rosstandarta ot 31.01.2014 N 14-st) (red. ot 12.08.2021). (in Russian)
5. Kamath C., Bukhari S., Dengel A. Comparative Study between Traditional Machine Learning and Deep Learning Approaches for Text Classification. Proceedings of the ACM Symposium on Document Engineering. 2018.
6. Slavov S., Tagarev A., Tulechki N., Boytcheva S. Company Industry Classification with Neural and Attention-Based Learning Models. Big Data. Knowledge and Control Systems Engineering (BdKCSE). 2019. 1−7.
7. Wood S., Muthyala R., Jin Y., Qin Y., Rukadikar N., Gao H., Rai A. Automated Industry Classification with Deep Learning. IEEE 12th International Conference on Semantic Computing (ICSC). 2018. 64−70.
8. Husmann S., Shivarova A., Steinert R. Company classification using machine learning. Expert Systems with Applications. 2022. V. 195.
9. Bernstein A., Clearwater S., Provost F. The Relational Vector-Space Model. Proceedings of IJCAI Workshop on Statistical Models from Relational Data. 2003. P. 8−18.
10. Kakushadze Z., Yu W. Statistical Industry Classification. SSRN Electronic Journal. 2016. 10.2139/ssrn.2802753.
11. Dolphin R., Smyth B., Dong R. A Machine Learning Approach to Industry Classification in Financial Markets. Artificial Intelligence and Cognitive Science. AICS 2022. Communications in Computer and Information Science. V. 1662. Springer. Cham.
12. Cai H., Zheng V.W., Chang K.C. A Comprehensive Survey of Graph Embedding: Problems, Techniques and Applications. IEEE Transactions on Knowledge and Data Engineering. 2017. 30: 1616-1637.
13. Zhou J., Cui G. Graph neural networks: A review of methods and applications. arXiv preprint arXiv:1812.08434. 2018.
14. Zhang S., Tong H., Xu J., Maciejewski R. Graph convolutional networks: a comprehensive review. Computational Social Networks. 2019. 6:1. doi:10.1186/s40649-019-0069-y.
15. Mikolov T., Chen K., Corrado G., Dean J. Efficient estimation of word representations in vector space. Proceedings of the International Conference on Learning Representations. 2013.
16. Kipf T., Welling M. Semi-supervised classification with graph convolutional networks. International Conference on Learning Representations (ICLR). 2017.
17. Hamilton W., Ying Z., Leskovec J. Inductive representation learning on large graphs. Proc. of NIPS. 2017. 1024–1034.
18. Yang B., Yih W., He X., Gao J., Deng L. Embedding Entities and Relations for Learning and Inference in Knowledge Bases. CoRR, abs/1412.6575. 2014.
19. Prikaz Minfina Rossii ot 20.07.2010 № 66n "O formakh bukhgalterskoi otchetnosti organizatsii". (in Russian)
20. Maaten L.V., Hinton G.E. Visualizing Data using t-SNE. Journal of Machine Learning Research. 2008. 9: 2579-2605.