IMPROVING THE QUALITY OF RECOMMENDATION SYSTEMS BASED ON QUALIMETRIC MEASUREMENT METHODS

Abstract

In the work on the basis of the conducted analysis of methods of carrying out scientific researches it is defined that the method of multifactor experiment most thoroughly allows to investigate influence of simultaneously various external factors on information system. This method is comprehensive and allows to obtain statistics for different testing scenarios, to determine the allowable limits of the impact of various external factors on the information system and their various aggregates, as well as to develop a set of recommendations for operation and maintenance of the information system. This article examines the relevance of the topic of referral systems in 2022, along with the growth of e-commerce, in which referral systems are most often used. The types of recommendation systems are analyzed on the basis and the shortcomings of each of them are considered in detail: collaborative filtering; content-based; knowledge-based; hybrid (hybrid).

The purpose of the research is formed, which is to improve the quality and systematization of quality parameters of recommendation systems on the basis of qualimetric methods and means of obtaining indicators of recommendation systems based on analytical tools. There is a scientific substantiation of the introduction of additional parameters to the algorithm of recommendations based on the "opinion" of the user and the introduction of the qualimetric method "Cyclogram of the quality of the recommendation system".

A scientific task has been formed, which includes: systematization of quality indicators of recommendation systems on the basis of qualimetric methods; development of new quality indicators of recommendation systems based on qualimetric methods; development of a new recommendation system of collaborative filtering, which includes assessment of the "opinion" of the user; improving the quality of the recommendation system by adding additional qualitative quality indicators to the algorithm of recommendations.

The expediency of using qualimetric methods for assessing the quality of recommendation systems is described. The importance of analyzing car reviews is described. The parameters of qualimetry and the problems we solve with the help of these parameters in the cyclogram of quality are considered.

In this scientific work a new qualimetric approach to measuring the quality of recommendation systems is considered. A system for assessing the quality of feedback and a recommendation system in the format of "Quality Cyclograms" is proposed, which includes the parameter of assessing the user's opinion about the product - "Value of Mids". The set of all eight indicators of the quality of recommendation systems is divided into key categories. The importance of research for business tasks is described. Conclusions and final conclusions on scientific work are formed.