Detection of bots in social networks by means of multilayered perceptron

Authors: Khachatryan M.G., Chepik P.I.
Published in issue: #4(33)/2019
DOI: 10.18698/2541-8009-2019-4-463
Category: Informatics, Computer Engineering and Control | Chapter: Methods and Systems of Information Protection, Information Security
Keywords: social networks, Twitter, bot, account, neural networks, cross-validation, metric, classification
Published: 09.04.2019

The possibility of using multilayer perceptron for the detection of bots in social networks is considered. The purpose of the study is to quantify the quality of detection of bots by a multilayer perceptron. Using the cross-validation algorithm, we determined the optimal hyperparameters for the multilayer perceptron and the value of the quality of classification for the found hyperparameters. The training and testing of the multilayer perceptron was carried out on the basis of a sample of several thousand Twitter accounts, consisting of both real users and bots of two different types. As a result of testing on this data the metric value is obtained = 0,958.

References

[1] Alymov A.S., Baranyuk V.V., Smirnova O.S. Detection of bot programs that mimic the behavior of people in the social network "Vkontakte". International Journal of Open Information Technologies, 2016, vol. 4, no. 8, pp. 55–60 (in Russ.).

[2] Gayo-Avello D. Social media won’t free us. IEEE Internet Comput., 2017, vol. 21, no. 4, pp. 98–101. DOI: 10.1109/MIC.2017.2911439 URL: https://ieeexplore.ieee.org/document/7994552

[3] Lyfenko N.D. Virtual users in social networks: myths and realities. Voprosy kiberbezopasnosti [Voprosy kiberbezopasnosti], 2014, no. 5(8), pp. 17–20 (in Russ.).

[4] Rosenblatt F. The perceptron: a probabilistic model for information storage and organization in the brain. Psychol. Rev., 1958, vol. 65, no. 6, pp. 386–408. DOI: 10.1037/h0042519 URL: https://psycnet.apa.org/doiLanding?doi=10.1037%2Fh0042519

[5] Rychagov M.N. Neuron networks: multilayer perceptron and Hopfield nets. Exponenta Pro. Matematika v prilozheniyakh, 2003, no. 1, pp. 29–37 (in Russ.).

[6] Xu B., Wang N., Chen T., et al. Empirical evaluation of rectified activations in convolution network. URL: https://arxiv.org/abs/1505.00853 (accessed: 15.01.2019).

[7] Yonaba H., Anctil F., Fortin V. Comparing sigmoid transfer functions for neural network multistep ahead streamflow forecasting. J. Hydrol. Eng., 2010, vol. 15, no. 4. DOI: 10.1061/(ASCE)HE.1943-5584.0000188 URL: https://ascelibrary.org/doi/10.1061/%28ASCE%29HE.1943-5584.0000188

[8] Kingma D., Ba J.L. Adam: a method for stochastic optimization. Proc. ICLR, 2015. URL: https://arxiv.org/abs/1412.6980 (accessed: 15.01.2019).

[9] Cresci S., Pietro R.D., Petrocchi M., et al. The paradigm-shift of social spambots: evidence, theories, and tools for the arms race. Proc. 26th Int. Conf. World Wide Web Companion, 2017, pp. 963–972. DOI: 10.1145/3041021.3055135 URL: https://dl.acm.org/citation.cfm?doid=3041021.3055135

[10] Raschka S. Model evaluation, model selection, and algorithm selection in machine learning. URL: https://sebastianraschka.com/pdf/manuscripts/model-eval.pdf (accessed: 15.01.2019).