Target-Aspect-Sentiment Detection on Hotels
71.47F1 ScoreLLM
Evaluation Results
| Method | Links | |
|---|---|---|
| LLMMode=Fine-tuned2026.05 | 71.47 | |
| MvPMode=Fine-tuned2026.05 | 69.37 | |
| LLM-MvP# Shots=1002026.05 | 68.38 | |
| Para.Mode=Fine-tuned2026.05 | 67.7 | |
| LLM-MvP# Shots=502026.05 | 67.42 | |
| Self-consistency# Shots=1002026.05 | 66.92 | |
| Single-order# Shots=1002026.05 | 66.06 | |
| Self-consistency# Shots=502026.05 | 65.03 | |
| Single-order# Shots=502026.05 | 64.44 | |
| LLM-MvP# Shots=102026.05 | 64.33 | |
| Self-consistency# Shots=102026.05 | 61.95 | |
| Single-order# Shots=102026.05 | 61.09 | |
| LLM-MvP# Shots=Avg.2026.05 | 59.3 | |
| Self-consistency# Shots=Avg.2026.05 | 57.72 | |
| Reasoning# Shots=502026.05 | 57.26 | |
| Reasoning# Shots=1002026.05 | 56.75 | |
| Gemma-3-27B ICLScenario=10 annotated examples, # Train=102026.03 | 56.51 | |
| Single-order# Shots=Avg.2026.05 | 56.38 | |
| LA-ABSA w/ ParaphraseScenario=10 annotated examples, # Train=full2026.03 | 55.69 | |
| LA-ABSA w/ DLOScenario=10 annotated examples, # Train=full2026.03 | 55.27 | |
| Reasoning# Shots=102026.05 | 53.88 | |
| Reasoning# Shots=Avg.2026.05 | 53 | |
| Reasoning# Shots=02026.05 | 44.09 | |
| LA-ABSA w/ ParaphraseScenario=0 annotated examples, # Train=full2026.03 | 40.74 | |
| LA-ABSA w/ DLOScenario=0 annotated examples, # Train=full2026.03 | 40.7 | |
| Gemma-3-27B (0-shot)Scenario=0 annotated examples, # Train=02026.03 | 38.97 | |
| LLM-MvP# Shots=02026.05 | 37.07 | |
| Self-consistency# Shots=02026.05 | 36.97 | |
| Single-order# Shots=02026.05 | 33.93 |