Multi-entity Reasoning on MEBench Set2 (11-100)
95.2Comparison AccuracyDocSage
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DocSage2026.03 | 95.2 | 92.3 | 81.8 | 90.6 | |
| StructRAG2026.03 | 79.3 | 60.1 | 65.7 | 67.6 | |
| GPT-4o + RAGRAG=true2026.03 | 77.7 | 61.3 | 66.7 | 67.9 | |
| GraphRAG2026.03 | 71.4 | 58.9 | 70.7 | 65.9 | |
| GPT-4oRAG=false2026.03 | 38.8 | 50.5 | 52.5 | 47.3 |