A comprehensive literature review on phishing URL detection using deep learning techniques

ABSTRACT

Strong and efficient defences have to be developed in response to the more-sophisticated phishing attempts is deep learning algorithms. The extensive analysis, which spans 41 research studies from 2019 to 2024, examines the state-of-the-art in deep learning for phishing URL identification. The review groups the studies according to feature engineering techniques such as character-level representations, word embeddings, and handcrafted features, and deep learning model architectures such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs, LSTMs), and hybrid models. To evaluate the relative advantages of various approaches, quantitative comparisons of stated performance metrics such as accuracy, precision, recall, and F1-score are offered. Robustness against adversarial assaults, real-time deployment and integration, extensive evaluation criteria, and the requirement for interpretability and explainability are some of the major issues and constraints. Promising avenues for further research are highlighted, including multimodal tactics, adversarial training, explainable AI techniques, zero-day attack detection, effective real-time deployment strategies, and large-scale assessment benchmarks offering a thorough understanding of the current situation and opens the door for future research into creating more reliable, understandable, and deployable deep learning solutions to counter the ever-evolving threat of phishing attacks. It does this by synthesising the most recent advancements and highlighting crucial gaps.

KEYWORDS:

• Uniform resource locator (URL)
• deep learning
• phishing
• threat actor

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported that there is no funding associated with the work featured in this article.