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

Table 9

Comparative analysis between solutions proposed in the present study and other studies.

Solution	Based perspective	Accuracy	Error rate	Advantage	Disadvantage
Cantina [60]	Content and identity (static analysis)	95%	5%	Fast; Can be integrated with phishing toolbars	Information gathered is still reduced; Knowledge about URL is limited
Daeef et al. [61]	Lexical and machine learning (hybrid analysis)	92.24%	5.40%	Wide scope and fast phishing detection system	High false positive rate
Yang, Zhao, Zen. [62]	Blacklist, lexical, and deep learning CNN (static analysis)	98.99%	0.59%	Fast; Based on deep learning	Needs improvement and more features
Jain and Gupta [63]	Visual similarity and machine learning (static analysis)	99.72%	1.89%	Fast to recognize targeted victims	Limited to e-banking websites; Knowledge about URL is limited
Solution provided in the existing study	Blacklist; Lexical; Content; Identity; Visual similarity; Behavioral; Machine or deep learning (hybrid analysis)	97.94%	2.1%	Fast; Based on rules; Trusted; Complete knowledge about a URL	Time and resource consuming when the whole process should be performed