METHOD OF MARKETPLACE LEGITIMATE USER AND ATTACKER PROFILING

Author:

Tsyrkaniuk Diana1ORCID,Sokolov Volodymyr1ORCID,Mazur Nataliia1ORCID,Kozachok Valerii1ORCID,Astapenya Volodymyr1ORCID

Affiliation:

1. Borys Grinchenko Kyiv University

Abstract

The number and complexity of cybercrime are constantly growing. New types of attacks and competition are emerging. The number of systems is growing faster than new cybersecurity professionals are learning, making it increasingly difficult to track users' actions in real-time manually. E-commerce is incredibly active. Not all retailers have enough resources to maintain their online stores, so they are forced to work with intermediaries. Unique trading platforms increasingly perform the role of intermediaries with their electronic catalogs (showcases), payment and logistics services, quality control - marketplaces. The article considers the problem of protecting the personal data of marketplace users. The article aims to develop a mathematical behavior model to increase the protection of the user's data to counter fraud (antifraud). Profiling can be built in two directions: profiling a legitimate user and an attacker (profitability and scoring issues are beyond the scope of this study). User profiling is based on typical behavior, amounts, and quantities of goods, the speed of filling the electronic cart, the number of refusals and returns, etc. A proprietary model for profiling user behavior based on the Python programming language and the Scikit-learn library using the method of random forest, linear regression, and decision tree was proposed, metrics were used using an error matrix, and algorithms were evaluated. As a result of comparing the evaluation of these algorithms of three methods, the linear regression method showed the best results: A is 98.60%, P is 0.01%, R is 0.54%, F is 0.33%. 2% of violators have been correctly identified, which positively affects the protection of personal data.

Publisher

Borys Grinchenko Kyiv University

Subject

General Medicine

Cited by 2 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3