Decision making by applying machine learning techniques to mitigate spam SMS attacks

Hisham AbouGrad*, Salem Chakhar, Ahmed Abubahia

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Downloads (Pure)

Abstract

Due to exponential developments in communication networks and computer technologies, spammers being provided with more options and tools to deliver their spam SMS attacks. This makes spam mitigation area as one of the hottest research areas in recent years, and also, affects privacy and causes revenue loss. Therefore, tools for making accurate decision whether spam or not are in compelling need. In this paper, spam mitigation method is proposed for recognising spam from authentic (non-spam) and the different processes used to mitigate spam SMS attacks. Also, anti-spam measures applied for classifying spam SMS with the aim to achieve higher classification accuracy throughout different classification methods to find the best classifier model. This paper contributes by giving the appropriate machine learning (ML) techniques with the use of decision-making paradigms to formulate a solution model, which can be applied for mitigating spam SMS attacks. The proposed model combines both ML techniques and the Delphi decision-making method. First, K-means clustering implemented to group words based on their occurrence similarities. Thereafter, text clustering applied with a variety of cluster numbers (10, 20 and 30 clusters respectively). Lastly, three different classifiers selected to cluster the dataset, which are Naive Bayes, Support Vector Machine and Random Forests. These classifiers are re-nowned as easy to use, efficient and more accurate in comparison with other classifiers. The research findings indicated clearly that the number of clusters combined with the number of attributes (columns) have revealed a significant influence on the classification accuracy performance.
Original languageEnglish
Title of host publicationKey Digital Trends in Artificial Intelligence and Robotics
Subtitle of host publicationProceedings of 4th International Conference on Deep Learning, Artificial Intelligence and Robotics, (ICDLAIR) 2022 - Progress in Algorithms and Applications of Deep Learning
EditorsLuigi Troiano, Alfredo Vaccaro, Nishtha Kesswani, Irene Díaz Rodriguez, Imene Brigui, David Pastor-Escuredo
PublisherSpringer
Pages154-166
Number of pages13
ISBN (Electronic)9783031303968
ISBN (Print)9783031303951
DOIs
Publication statusPublished - 17 Apr 2023

Publication series

NameLecture Notes in Networks and Systems
PublisherSpringer
Volume670
ISSN (Print)2367-3370
ISSN (Electronic)2367-3389

Keywords

  • machine learning algorithms
  • feature classification algorithms
  • decision-making method
  • mitigating spam techniques
  • spam analytics model
  • mobile network security
  • privacy solution

Cite this