WebShell Detection on 26.59K PHP scripts dataset (test)
99.75AccuracyGPT-4
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-4Category=LLM + BFAD, Enhanced By=BFAD2025.04 | 99.75 | 100 | 98.71 | 99.35 | |
| LLaMA-3.1-70BCategory=LLM + BFAD, Enhanced By=BFAD2025.04 | 99.38 | 98.72 | 98.09 | 98.4 | |
| Qwen-2.5-Coder-14BCategory=LLM + BFAD, Enhanced By=BFAD2025.04 | 98.76 | 98.68 | 94.9 | 96.75 | |
| Qwen-2.5-Coder-14BCategory=LLM Baselines (Large)2025.04 | 98.64 | 99.32 | 93.63 | 96.39 | |
| GATCategory=Graph Baselines2025.04 | 98.37 | 99.52 | 97.39 | 98.87 | |
| LLaMA-3.1-70BCategory=LLM Baselines (Large)2025.04 | 98.01 | 97.31 | 92.36 | 94.77 | |
| Qwen-2.5-3BCategory=LLM + BFAD, Enhanced By=BFAD2025.04 | 97.39 | 88.64 | 99.36 | 93.69 | |
| GPT-4Category=LLM Baselines (Large)2025.04 | 97.27 | 100 | 85.98 | 92.46 | |
| GCNCategory=Graph Baselines2025.04 | 96.9 | 94.4 | 95.3 | 94.9 | |
| CodeBERT+RFCategory=Sequence Baselines2025.04 | 96.3 | 94 | 95.6 | 94.8 | |
| GloVe+SVMCategory=Sequence Baselines2025.04 | 96.2 | 93.3 | 94.3 | 93.8 | |
| Qwen-2.5-3BCategory=LLM Baselines (Small)2025.04 | 93.72 | 78.03 | 91.84 | 84.37 | |
| Qwen-2.5-0.5BCategory=LLM + BFAD, Enhanced By=BFAD2025.04 | 91.94 | 71.1 | 98.73 | 82.67 | |
| Qwen-2.5-1.5BCategory=LLM + BFAD, Enhanced By=BFAD2025.04 | 80.4 | 48.51 | 100 | 65.33 | |
| Qwen-2.5-Coder-3BCategory=LLM + BFAD, Enhanced By=BFAD2025.04 | 78.89 | 46.67 | 100 | 63.64 | |
| Qwen-2.5-Coder-3BCategory=LLM Baselines (Small)2025.04 | 71.11 | 38.93 | 99.32 | 55.93 | |
| Qwen-2.5-1.5BCategory=LLM Baselines (Small)2025.04 | 43.62 | 34.61 | 95.77 | 50.84 | |
| Qwen-2.5-0.5BCategory=LLM Baselines (Small)2025.04 | 19.47 | 18.65 | 100 | 31.44 |