Trace Reconstruction on APT-Eval-Trace D4
82.7RecallHunterAgent
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| HunterAgentMethodology Category=LLM & agentic baselines, Backbone=GPT-4o, Log-suppression rate=30%2026.05 | 82.7 | 91.3 | 86.8 | 6.4 | |
| Tool-Aug. ReActMethodology Category=LLM & agentic baselines, Backbone=GPT-4o, Log-suppression rate=30%2026.05 | 78.4 | 46.2 | 58.1 | 61.5 | |
| HunterAgent†Methodology Category=LLM & agentic baselines, Backbone=GPT-4o, Log-suppression rate=30%2026.05 | 76.1 | 64.8 | 70 | 28.6 | |
| KAIROSMethodology Category=Recent learned provenance / streaming, Log-suppression rate=30%2026.05 | 60.4 | 73.9 | 66.5 | 14.8 | |
| GraphRAGMethodology Category=LLM & agentic baselines, Backbone=GPT-4o, Log-suppression rate=30%2026.05 | 58.2 | 31.7 | 41 | 74.1 | |
| NODLINKMethodology Category=Recent learned provenance / streaming, Log-suppression rate=30%2026.05 | 53.8 | 79.2 | 64.1 | 9 | |
| GNN-LinkPredictMethodology Category=Statistical relational learning, Log-suppression rate=30%2026.05 | 43.1 | 68.5 | 52.9 | 36.8 | |
| Zero-Shot LLMMethodology Category=LLM & agentic baselines, Backbone=GPT-4o, Log-suppression rate=30%2026.05 | 34.6 | 12.4 | 18.2 | 92.3 | |
| MAGICMethodology Category=Deterministic provenance graphs, Log-suppression rate=30%2026.05 | 26.2 | 98.8 | 41.4 | 0.4 | |
| SLEUTHMethodology Category=Deterministic provenance graphs, Log-suppression rate=30%2026.05 | 18.5 | 99.1 | 31.2 | 0 |