from 01.01.2022 to 01.01.2024
Krasnoyarsk, Krasnoyarsk, Russian Federation
Krasnoyarsk, Krasnoyarsk, Russian Federation
Recommendation systems are widely used in online services to provide personalized recommendations to users, which helps to improve the user experience and increase business efficiency. This article provides a comparative analysis of the mathematical algorithms used to build recommendation systems. The article discusses the main classes of algorithms, such as collaborative filtering, content methods, hybrid approaches and algorithms based on matrix decompositions, as well as clustering methods. For each class of algorithms, there is an overview of the basic principles of research, advantages and disadvantages. A comparative analysis of algorithms based on their accuracy, speed, scalability and ability to work with sparse data is carried out. The issue of resistance of algorithms to cold starts and ways to overcome it are also considered. Based on the analysis, it is concluded that there is no universal algorithm that would be suitable for all types of recommender system problems. The selection of the optimal algorithm depends on the specific task, data characteristics and requirements for the quality of recommendations. Further research in this area can be aimed at developing new hybrid algorithms that take into account the features of modern online services and improve the quality of recommendations, as well as the development of recommender systems built on the basis of these algorithms.
recommendation systems, collaborative filtering, clustering, estimation prediction, user similarity, subject similarity, Python, SVD decomposition
1. Tsurko, V.V. (2019). Recommender systems in healthcare. Managing large systems, 82, 61–73.
2. Recommender systems [Electronic source] // ITMO Wikinotes. – 2022. – URL: https://neerc.ifmo.ru/wiki/index.php?title=%D0%A0%D0%B5%D0%BA%D0%BE%D0%BC%D0%B5%D0%BD%D0%B4%D0%B0%D1%82%D0%B5%D0%BB%D1%8C%D0%BD%D1%8B%D0%B5_%D1%81%D0%B8%D1%81%D1%82%D0%B5%D0%BC%D1%8B (Date of access: 11.06.2024).
3. Yakupov, D. R., Namiot, D. E. (2022). Session-based recommender systems – models and tasks. International Journal of Open Information Technologies, 7(10), 128–152.
4. Parygin, D. S., Strekalova, A. S., Gurtyakov, A. S., et. al. (2019). Application of recommendation technologies in systems with spatial information. Caspian Journal: Management and High Technologies, 1(45), 96–109.
5. Filippov, S. A., Zakharov, V. N. (2016). Assessing the effectiveness of a method for increasing the persistence of information in life support recommendation systems based on implicit data. Modern information technologies and IT education, 2(12), 186–193.
6. Larionov, V. S., Dunin, I. V. (2017). Review of collaborative filtering methods. Forum of young scientists, 5(9), 1230–1238.
7. Ershov, T. A. (2023). Recommendation system for images based on artificial intelligence. E-Scio, 8(83), 1–7.
8. What are recommender systems and how do they work? Skillbox Media. [Electronic source] URL: https://skillbox.ru/media/code/chto_takoe_rekomendatelnye_sistemy_i_kak_oni_rabotayut/ (Date of access: 10.06.2024).
9. Recommender systems: what they are and how the recommendation algorithm works [Electronic source] URL: https://mindbox.ru/journal/education/rekomendatelnye-sistemy/ (Date of access: 27.05.2024).
10. Recommendation systems: how algorithms select products and content for us. [Electronic source]. URL: https://hub.forklog.com/rekomendatelnye-sistemy-kak-algoritmy-podbirayut-nam-tovary-i-kontent/ (Date of access: 30.05.2024).
11. What are recommender systems? [Electronic source] URL: https://forklog.com/cryptorium/ai/chto-takoe-rekomendatelnye-sistemy (Date of access: 01.06.2024).
12. Koroleva, D. E., Filippov, M. V. (2013). Analysis of learning algorithms for collaborative recommender systems. Engineering journal: science and innovation, 6(18), 1–8.
13. Kim, T. H., Yang, S. B. (2007). A Refined Neighbor Selection Algorithm for Clustering-Based Collaborative Filtering. The KIPS Transactions: PartD, 14(3), 347–354.
14. MovieLens. [Electronic source] URL: https://grouplens.org/datasets/movielens/ (Date of access: 30.05.2024).
15. How recommender systems work. [Electronic source] URL: https://neurohive.io/ru/osnovy-data-science/rekomendatelnye-sistemy-modeli-i-ocenka/ (Date of access: 10.06.2024).
16. 15 libraries Python for Data Science. [Electronic source] URL: https://blog.skillfactory.ru/15-bibliotek-python-dlya-data-science/ (Date of access: 12.06.2024).
17. Hastie, T., Mazumder R., Lee, J. D. (2015). Matrix completion and low-rank SVD via fast alternating least squares. The Journal of Machine Learning Research, 16(1), 3367–3402.
18. Takács, G., Tikk, D. (2022). Alternating least squares for personalized ranking. Proceedings of the 6th ACM conference on Recommender systems, 83–90.
19. Deng, D. (2020). DBSCAN clustering algorithm based on density. Proceedings of 7th international forum on electrical engineering and automation, 949–953.
20. Zhilov, R. A. (2023). Intelligent methods of data clustering. News of the KBSC RAS, 6(116), 152–159.
21. Tokuda, E. K., Comin, C. H., Costa, L. D. F. (2022). Revisiting agglomerative clustering. Physica A: Statistical mechanics and its applications, 585, 126433.
22. Bulyga, F. S., Kureichik, V. M. (2021). Agglomerative clustering algorithms in relation to problems of analysis of linguistic expert information. Izvestia SFU. Technical science, 6(223), 73–88.
23. Evaluation of clustering performance in Scikit Learn. [Electronic source] URL: https://ru.linux-console.net/?p=25753 (Date of access: 12.06.2024).
24. Taratukhina, Yu. V., Bart, T. V., Vlasov, V. V. (2019). Machine learning model of an information recommendation system on issues of individualization of education. Educational resources and technologies, 2(27), 7–14.
25. Sharma, M., Mittal, R., Bharati, A., et. al. (2022). A Survey and Classification on Recommendation Systems. Proceedings of 2nd International Conference on Big Data, Machine Learning and Applications, 1–16.
