The Information-Analytical Bots Detection System Based On The Assembly Of Classifiers

EasyChair Preprint 4517

Abstract

Currently, the use of bots, disguised as ordinary users of social networks and guidance with special programs, has serious consequences. For example, bots were used to influence political elections, distort information on the Internet, and manipulate stock prices on the stock exchange. The detection of bots in social networks is carried out by many research teams, the areas of research of which include the use of machine learning methods. However, the practical results of detecting bots on social networks indicate significant limitations, since the methodological tools used have language limitations and ineffective criteria for determining bots. The report provides a description of the information-analytical system (client-server application) that allows for the collection and analysis of data of social networks in order to identify bots. The application is based on a bots detection module based on the assembly of classifiers of social network accounts, the capabilities of which allowed to minimize the risk of bot detection errors. The practical utilizing of the application allows increasing the operatively and effectiveness of detecting bots in comparison with other approaches.

Keyphrases: Ensemble of Models, association of classifiers, bots detection, machine learning, social networks

@booklet{EasyChair:4517,
  author    = {Vladimir Nikiforovich Kuzmin and Artem Bakitzhanovich Menisov and Ivan Anatolevich Shastun},
  title     = {The Information-Analytical Bots Detection System Based On The Assembly Of Classifiers},
  howpublished = {EasyChair Preprint 4517},
  year      = {EasyChair, 2020}}