DSmishSMS-A System to Detect Smishing SMS.

Area of Science:

• Cybersecurity and Information Security
• Machine Learning Applications
• Mobile Device Security

Background:

• Smartphones are integral to daily life, increasing vulnerability to targeted attacks like smishing (SMS Phishing).
• Smishing attacks leverage text messages, posing unique detection challenges due to limited information, abbreviations, and symbolic language.
• Scarcity of real-world smishing datasets complicates the development of effective detection models.

Purpose of the Study:

• To develop and evaluate an efficient smishing detection model.
• To address the challenges of detecting smishing attacks with limited features and data.
• To enhance the security of smartphone users against SMS Phishing threats.

Main Methods:

• A two-phase smishing detection model was proposed: Domain Checking Phase and SMS Classification Phase.
• The Domain Checking Phase scrutinizes the authenticity of Uniform Resource Locators (URLs) within SMS messages.
• The SMS Classification Phase extracts key features from message content and employs the Backpropagation Algorithm for classification, comparing results with traditional classifiers.

Main Results:

• The proposed smishing detection model achieved a high accuracy of 97.93% in evaluations.
• The system effectively utilizes a limited set of five extracted features for machine learning classification.
• The two-phase approach, integrating URL authenticity and content analysis, proved highly efficient.

Conclusions:

• The developed smishing detection model demonstrates significant efficiency and accuracy.
• The integration of domain checking and SMS content analysis is crucial for robust SMS Phishing detection.
• The proposed method offers a viable solution for combating the growing threat of smishing attacks on smartphones.