Hybrid Rule-Based Solution for Phishing URL Detection Using Convolutional Neural Network

Table 3

Comparison of advantages and disadvantages of different phishing detection techniques.

Methods	Advantages	Disadvantages
Blacklist	Does not require high resources to use it; Fast; Excellent when minimal FP (false positive) required	Not trusted; Not always updated
Lexical and host	Provides a sort of lexical profile for such URL; Can prevent and not only detect	Should always be updated; Needs human intervention sometimes
Content	Trusted; Detect hidden content like iframes	Cannot detect obfuscation; Should be maintained
Identity	Offers an owner profile	Not trusted
Visual similarity	Effective; Mitigate zero-hour attack	Higher FP rate; High computational cost
Behavioral	Detects hidden anomalies; Reveals novel abnormal behaviors	Very high computational cost; Needs human intervention; Should be maintained
Machine learning	Can prevent and not only detect; Mitigate zero-day attack; Construct its own models	Time-consuming; Needs maintenance; Works through several rules
Deep learning	Can prevent and not only detect; Mitigate zero-day attack; Constructs its own models; Becomes more effective with data increase	Time-consuming; Should be maintained; A vast number of rules; Massive number of parameters to handle