METHODS FOR DEVELOPING RECOMMENDATION SYSTEMS
DOI:
https://doi.org/10.31649/1999-9941-2021-52-3-10-15Keywords:
recommendation system, cold start problem, web service, machine learning, algorithmsAbstract
The basic principles of building a recommendation system and methods for solving the problem of cold start arising from insufficient interaction of the user with the software at the initial stages of working with it are considered. The efficiency of the recommender system has been increased when there is insufficient data sampling and when new elements appear in the system for which there are no statistics.
References
K. Falk, Practical recommender systems, 2020, 448 s. [in Russian].
Li L., Chu W., Langford J., Schapire R. E., «A contextual-bandit approach to personalized news arti-cle recommendation», Proceedings of the 19th International Conference on World Wide Web. р. 661–670. 2010.
D. Bugaychenko, A. Dzuba, «Musical recommendations and personalization in a social network», RecSys '13: Proceedings of the 7th ACM conference on Recommender systems, October 2013, p. 367–370.
Auer P., Cesa-Bianchi N., Fischer P., «Finite-time analysis of the multiarmed bandit problem», Machine Learning., vol. 47, no 2–3, p. 235–256. 2002.
Yahoo! Front page today module user click log dataset, version 1.0 (1.1 GB). [Online]. Available: https://webscope.sandbox.yahoo.com/catalog.php?datatype=r&did=49. Accessed Nov. 15, 2021.
R. Z. Omarov, A. V. Vostrotina, A. D. Li, «Problema "kholodnoho startu"», Molodyi uchenyi, № 26 (264), s. 85-88. 2019 [in Russian].
Downloads
-
PDF (Українська)
Downloads: 546
