Phishing Detection on Mendeley and TR-OP averaged
90.57PrecisionPhishDebate
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PhishDebateBase LLM=GPT-4o-mini2025.06 | 90.57 | 93.9 | 98 | 94.14 | 37.5 | |
| CoTBase LLM=GPT-4o-mini2025.06 | 88.61 | 90.7 | 93.4 | 90.94 | 10.5 | |
| Single AgentBase LLM=GPT-4o-mini2025.06 | 60.57 | 67 | 97.4 | 74.69 | 4.7 |