Scholarlink

Email spamming is one of the most common scams that are occurring nowadays. In this method, an email, seemingly sent from a trusted address or sender, can be used for various purposes, including advertisement, commercial and non-commercial intents. Furthermore, it may contain content harmful to the recipient, a tactic known as phishing. Despite several email spam blocking mechanisms employed by the email service providers, it is still a major issue to handle. In this paper, several machine learning algorithms have been investigated over a realistic dataset including (but not limited to) Naïve Bayes (NB), Support Vector Machine (SVM), and Decision Tree (DT) classifiers. Among them Naïve Bayes classifier exhibited the highest accuracy at 99.8% followed by AdaBoost at 96.7%. Analyses showed scheme was promising in terms of accuracy and it outperformed various state-of-the-art approaches in the literature upon comparison.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Email Spam Classification: A Machine Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Atta Rahman,
  • Linah Saraireh,
  • Mustafa Youldash,
  • Wafa Hantom,
  • Dania Alkhualifi,
  • Majed Nabil,
  • Ashraf Saadeldeen,
  • Maqsood Mahmud,
  • Asiya Abdus Salam,
  • Mohammed Salih Ahmed,
  • Mohammed Gollapalli

摘要

Email spamming is one of the most common scams that are occurring nowadays. In this method, an email, seemingly sent from a trusted address or sender, can be used for various purposes, including advertisement, commercial and non-commercial intents. Furthermore, it may contain content harmful to the recipient, a tactic known as phishing. Despite several email spam blocking mechanisms employed by the email service providers, it is still a major issue to handle. In this paper, several machine learning algorithms have been investigated over a realistic dataset including (but not limited to) Naïve Bayes (NB), Support Vector Machine (SVM), and Decision Tree (DT) classifiers. Among them Naïve Bayes classifier exhibited the highest accuracy at 99.8% followed by AdaBoost at 96.7%. Analyses showed scheme was promising in terms of accuracy and it outperformed various state-of-the-art approaches in the literature upon comparison.