Fuzzy Systems improve the accuracy of machine learning detection of socialbots

  • Carla C. Pacheco IME
  • Alex Garcia IME
  • Raphael Machado Inmetro
  • Ronaldo M. Salles IME
DOI: https://doi.org/10.5753/sbsi.2019.7446

Abstract


Machine learning has been widely used in the detection of socialbots in Online Social Networks. This paper presents the use of an algorithm committee to improve the accuracy of socialbots identification. The committee combines the knowledge obtained by machine learning algorithms and human heuristic knowledge obtained through interviews and formalized in fuzzy rules. Results show that these approaches are complementary, since their use in a single committee presents accuracy above 93%, better than each of the algorithms independently.

Keywords: Social Networks, Detection of Socialbot, Machine Learning, Fuzzy Logic, Ensemble Learning

References

PACHECO, Carla de Castro; GARCIA, Alex de Vasconcellos; MACHADO, Raphael Carlos Santos; SALLES, Ronaldo Moreira. Sistemas fuzzy complementam a detecção de socialbots por aprendizado de máquina. Brasília: Revista Militar de Ciência e Tecnologia, 2018.
Published
2019-05-20
How to Cite

Select a Format
C. PACHECO, Carla; GARCIA, Alex; MACHADO, Raphael; M. SALLES, Ronaldo. Fuzzy Systems improve the accuracy of machine learning detection of socialbots. In: THESIS AND DISSERTATIONS ON INFORMATION SYSTEMS CONTEST - BRAZILIAN SYMPOSIUM ON INFORMATION SYSTEMS (SBSI), 15. , 2019, Aracaju. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2019 . p. 91-95. DOI: https://doi.org/10.5753/sbsi.2019.7446.