General Aspect Identification on Manually Annotated Reviews (test)
78.26Exact MatchLLM (Gemini)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| LLM (Gemini)Model Type=Large Language Model, Variant=Gemini2026.01 | 78.26 | 7.61 | 91.36 | 92.43 | |
| Rule-BasedApproach=Rule-Based system baseline2026.01 | 52.17 | 16.3 | 82 | 84 | |
| FT-RB DSBase Model=Gemma, Training Data=Rule-Based annotated dataset (RB-DS), Training Setup=Fine-Tuned2026.01 | 13.04 | 47.83 | 64.52 | 62.82 | |
| FT-LLM DSBase Model=Gemma, Training Data=Gemini-annotated dataset (LLM-DS), Training Setup=Fine-Tuned2026.01 | 8.7 | 55.43 | 60.47 | 58.88 |