Development of a method for searсhing for similar account In social networks based on cluster analysis

Abstract

In recent decades, the number of Internet users has been growing significantly. At the same time, the number of people using social networks to communicate and receive information is growing. The active development of social networks has also had a significant impact on the advertising market, as the information about users collected by these networks allows advertisers to target ads with great precision, which has a significant impact on the growth of sales of goods and services. After all, by identifying similar accounts, you can easily determine that a product that one user likes is likely to attract another. In order to select the most appropriate method of cluster analysis for the chosen topic, a systematic analysis of cluster analysis methods was carried out. It has been analyzed that one of the approaches to finding similar accounts in social networks is to analyze accounts by certain characteristics and is widely used by modern companies that own social networks for targeted advertising. The analysis of clustering methods showed that there is no specific universal method for analyzing data clustering, which is why various clustering methods were analyzed and it was concluded that the improved k-means algorithm — k-means mini batch is best suited for finding similar accounts.