Izvestiya Vysshikh Uchebnykh Zavedenii. Elektronika home

Publication of the journal

The section is currently being updated

Adaptation of contemporary approaches to local data specificity and user behavior makes it possible to improve efficiency of personalized recommendations. Currently the companies actively implement recommender system technologies to improve user recommendations and increase conversion. In this work, modern ranking algorithms in recommender systems are considered and an approach to their improvement is proposed. For the purposes of effectively forming up personal recommendations for new users having no sufficient history of communication with system objects, a mathematical model of object ranking is proposed, using a combined approach with account for historical data and contextual information. Algorithms for offline and online phases of recommendation forming-up have been developed. The offline algorithm involves collecting and analyzing a matrix of estimates and constructing a graph of implicit information. The online algorithm uses social and contextual data to select the optimal forecasting model. The modular architecture of the software package is implemented with the integration of additional data sources, such as OCR, GPT and external factors (weather, location, time of day).The effectiveness of the system was assessed by F1-measure metrics. According to the experimental results, the developed system showed an improvement quality metric of 2 and 12 % compared to Adobe Target and Amazon Personalize, respectively.

Key words: recommender systems, ranking algorithms, collaborative filtering, hybrid models, personalization, machine learning
Published in: INFORMATION-COMMUNICATION TECHNOLOGIES
Bibliography link: Bolotin Yu. S. Development of ranking algorithms with contextual adaptation for recommender systems. Izv.vuzov. Elektronika = Proc. Univ. Electronics. 2025;30(4):508–519. (InRuss.).https://doi.org/10.24151/1561-5405-2025-30-4-508-519

Yury S. Bolotin
National Research University of Electronic Technology (Russia, 124498, Moscow, Zelenograd, Shokin sq., 1)

1. Маркетинговое исследование. Интернет-торговля в России 2024. Data Insight. 03.05.2024. Available at: https://datainsight.ru/eCommerce_2023 (accessed: 30.05.2025).

Marketing research.E-commerce in Russia 2024.Data Insight.03.05.2024. (In Russ.). Availableat: https://datainsight.ru/eCommerce_2023 (accessed: 30.05.2025).

2. Воронцов К. В. Коллаборативная фильтрация и матричные разложения: курс лекций. Машинноеобучение. 21.03.2025. Available at: http://www.machinelearning.ru/wiki/images/9/95/Voron-ML-CF.pdf (accessed: 30.05.2025).

Vorontsov K. V. Collaborate filtering and matrix splittings: lecture course. Mashinnoe obuchenie = Machine Learning.(In Russ.). 21.03.2025. Available at: http://www.machinelearning.ru/wiki/images/9/95/Voron-ML-CF.pdf (accessed: 30.05.2025).

3. Falk K. Practical recommender systems. Shelter Island, NY: Manning Publ.; 2019. 432 p.

4. Shirokikh M., Shenbin I., Alekseev A., Volodkevich A., Vasilev A., Nikolenko S. User response modeling in recommender systems: A survey. Зап. науч. сем. ПОМИ. 2023;530:141–190. EDN: FWJUQZ.

Shirokikh M., Shenbin I., Alekseev A., Volodkevich A., Vasilev A., Nikolenko S. User response modeling in recommender systems: A survey. J. Math. Sci. 2024;285:255–293. https://doi.org/10.1007/s10958-024-07431-3

5. Jannach D., Zanker M., Felfernig A., Friedrich G. Recommender systems: An introduction. Cambridge, MA: CambridgeUniv. Press; 2020. 352 p.

6. Гагарина Л. Г., Болотин Ю. С., Болотина Е. С. Исследование и разработка методики фильтрации для рекомендательной системы. ИзвестияТулГУ. Техническиенауки. 2023;(1):387–390. https://doi.org/10.24412/2071-6168-2023-1-387-391. EDN: ECUYCW.

Gagarina L. G., Bolotin Yu. S., Bolotina E. S. Research and development of a filtering technique for a recommendation system.Izvestiya TulGU. Tekhnicheskie nauki = Proceedings of the TSU. Technical Sciences.2023;(1):387–390. (In Russ.). https://doi.org/10.24412/2071-6168-2023-1-387-391

7. Rossetti M., Stella F., Zanker M. Towards explaining latent factors with topic models in collaborative recommender systems. In: 2013 24th International Workshop on Database and Expert Systems Applications. LosAlamitos, CA: IEEE; 2013, pp. 162–167. https://doi.org/10.1109/DEXA.2013.26

8. Monastyrev V. V., Drobintsev P. D. Recommendation system based on user actions in the social network. Тр. Института системного программирования РАН = ProceedingsofISPRAS. 2020;32(3):101–108. https://doi.org/10.15514/ISPRAS-2020-32(3)-9

9. Болотина Е. С., Болотин Ю. С., Гагарина Л. Г. Разработка алгоритма первичной инициализации рекомендательных систем. Информационныесистемыитехнологии. 2025;(2):29–37. EDN: JOHJAQ.

Bolotina E. S., Bolotin Yu. S., Gagarina L. G. Development of algorithm for solving primary initialization problem in recommender system.Informatsionnye sistemy i tekhnologii = Information Systems and Technologies. 2025;(2):29–37. (In Russ.).

10. Ricci F., Rokach L., Shapira B. (eds). Recommender systems: handbook. 3^rded. New York: Springer; 2022. xi, 1060 p. https://doi.org/10.1007/978-1-0716-2197-4

Information about the publication

UDK: 004.85
Publication type: Scientific article
Source lang: Russian
DOI: 10.24151/1561-5405-2025-30-4-508-519
RISC id: XPQMFO

JATS XML