PhishSSL: Self-Supervised Contrastive Learning for Phishing Website Detection

2025-10-07

AI Summary

Phishing website detection faces challenges of high annotation costs and poor adaptability to emerging attacks. To address these, this paper proposes an unsupervised self-supervised contrastive learning framework that effectively identifies malicious websites without human-labeled data. Methodologically, it innovatively integrates hybrid tabular data augmentation with an adaptive feature attention mechanism, generating semantically consistent augmented views while enhancing discriminative feature representation. Evaluated on three heterogeneous datasets, the framework consistently outperforms state-of-the-art unsupervised and self-supervised methods. Ablation studies validate the individual contributions of each component. The approach demonstrates strong generalization capability, cross-dataset transferability, and robust stability under varying conditions. Overall, it establishes a novel paradigm for phishing detection in low-resource settings, advancing practical deployment where labeled data is scarce or costly to obtain.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Adversarial Attacks & Robustness
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Consent frameworks and practices on the web

Abstract

Phishing websites remain a persistent cybersecurity threat by mimicking legitimate sites to steal sensitive user information. Existing machine learning-based detection methods often rely on supervised learning with labeled data, which not only incurs substantial annotation costs but also limits adaptability to novel attack patterns. To address these challenges, we propose PhishSSL, a self-supervised contrastive learning framework that eliminates the need for labeled phishing data during training. PhishSSL combines hybrid tabular augmentation with adaptive feature attention to produce semantically consistent views and emphasize discriminative attributes. We evaluate PhishSSL on three phishing datasets with distinct feature compositions. Across all datasets, PhishSSL consistently outperforms unsupervised and self-supervised baselines, while ablation studies confirm the contribution of each component. Moreover, PhishSSL maintains robust performance despite the diversity of feature sets, highlighting its strong generalization and transferability. These results demonstrate that PhishSSL offers a promising solution for phishing website detection, particularly effective against evolving threats in dynamic Web environments.

Problem

• Detecting phishing websites without labeled training data
• Addressing high annotation costs and limited attack adaptability
• Improving generalization across diverse website feature sets

Innovation

• Self-supervised contrastive learning eliminates labeled data need
• Hybrid tabular augmentation creates semantically consistent views
• Adaptive feature attention emphasizes discriminative phishing attributes