COVID-19 malicious domain names classification.

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• The digital shift and COVID-19 pandemic increased vulnerability to cybercrime, particularly phishing attacks.
• Existing anti-phishing methods like blacklists and heuristics are insufficient against evolving threats.
• Cybercriminals exploit public interest in health crises for malicious activities.

Purpose of the Study:

• To develop and evaluate machine learning models for classifying COVID-19-related domain names.
• To identify effective features for distinguishing malicious from legitimate domains.
• To enhance cybersecurity defenses against pandemic-related phishing.

Main Methods:

• Utilized machine learning models for domain name classification.
• Extracted a limited set of lexical features from domain names.
• Incorporated the number of subdomain levels as a predictive feature.

Main Results:

• Machine learning models achieved high accuracy in classifying malicious COVID-19 domain names.
• A small set of carefully selected lexical features significantly improved model performance.
• The number of subdomain levels proved to be an influential feature in prediction.

Conclusions:

• Machine learning, using limited lexical features, offers an effective approach to combatting COVID-19-related phishing.
• Domain name analysis, including subdomain structure, is crucial for cybersecurity threat detection.
• Continuous innovation in cybersecurity is necessary to counter persistent cyber threats.