Authors: Yuexin Li, Chengyu Huang, Shumin Deng, Mei Lin Lock, Tri Cao, Nay Oo, Bryan Hooi, Hoon Wei Lim

Published on: March 04, 2024

Impact Score: 7.8

Arxiv code: Arxiv:2403.02253

Summary

• What is new: Introduction of an automated knowledge collection pipeline for phishing detection and the multimodal approach that uses both logos and textual information.
• Why this is important: Existing reference-based phishing detectors suffer from scalability issues and overlook the textual content on webpages, leading to false negatives.
• What the research proposes: Development of KnowPhish, a large-scale multimodal brand knowledge base, and KnowPhish Detector, a system that combines logo and text analysis for phishing detection.
• Results: KnowPhish and KnowPhish Detector have shown significant improvements in detecting phishing pages in both controlled evaluations and a field study, surpassing prior methods.

Technical Details

Technological frameworks used: Automated knowledge collection pipeline, Large Language Model (LLM) for text analysis

Models used: KnowPhish Detector (KPD)

Data used: KnowPhish brand knowledge base containing information on 20k brands

Potential Impact

Cybersecurity firms, financial institutions, and online retail companies could benefit from these insights by enhancing their phishing detection capabilities.

Want to implement this idea in a business?

We have generated a startup concept here: LogoGuard.