Publication: Detection of non-typical users of the electronic marketplace "freight transportation" to prevent the competitive intelligence
Дата
2021
Авторы
Journal Title
Journal ISSN
Volume Title
Издатель
Аннотация
© 2020 Elsevier B.V.. All rights reserved.This article presents results of usage of cluster analysis and machine learning to detect non-typical users of the electronic freight transportation marketplace. The most significant red flags to identify system users who collect commercial information have been identified, a competitor's profile has been drawn up. The use of various methods of clustering and classification in the modeling of this problem is studied. The procedure of formation of a training sample in the absence of labeled data has been carried out. The algorithm has been developed to detect non-typical users based on machine data analysis methods in order to prevent the real-time commercial intelligence. This algorithm is based on density-based spatial clustering of applications with noise (DBSCAN) and balanced iterative reducing and clustering using hierarchies (BIRCH) cluster analysis methods, identification of anomalous objects using the Isolation Forest method and ensemble Gradient Boosting algorithm with decision trees as the base classifiers. It can be used by electronic marketplace's owners, as well as by other companies that apply similar technologies while providing their services. As a result, software tool in Python has been developed, which allows to timely detect users collecting commercial information and hence to reduce a negative impact from possible competitive intelligence.
Описание
Ключевые слова
Цитирование
Domashova, J. Detection of non-typical users of the electronic marketplace "freight transportation" to prevent the competitive intelligence / Domashova, J., Zasypkina, A. // Procedia Computer Science. - 2021. - 190. - P. 210-216. - 10.1016/j.procs.2021.06.026