Analysis of Approaches to the Development of Recommendation Systems in the Field of Education
Zoya V. Arkhipova, Alexander Sorokin · System Analysis & Mathematical Modeling · 2024
The paper deals with the issues related to the development of recommendation systems in the field of educational services on high-tech digital platforms. The relevance of this problem is connected with the fact that recommendation systems, as well as methods and approaches to their creation, are constantly evolving, as they should adapt to the changing requirements of the educational services market and user preferences. Recently, recommendation systems based on neural networks have been actively used, the article analyzes the feasibility of using such systems in the field of educational services. Based on the analysis of existing methods and approaches to the development of recommendation systems, it is proposed to classify recommendation systems according to such parameters as: the type of data used; the method of training; the field of application; the complexity of the model; the degree of interaction with the user; the method of recommendation; the space of recommendations; the volume of recommendations. The problems arising in the process of development and application of recommendation systems are considered; it is proposed to unite the problems of recommendation systems into the following categories: the problem of models; the problem of limited data (cold start); the problem of filtering bubble; the problem of selecting metrics to assess the quality of the system; the problems of infrastructure and system efficiency; security problems; ethical problems. Based on the analysis and systematization of problems of use, approaches to their solution are proposed, and recommendations on the architecture of building recommendation systems used in the field of education are given.