Table 7 Model performance under different feature similarity calculation functions.
From: A multi-modal sarcasm detection model based on cue learning
Feature Similarity Calculation Function | ACC (%) | F1 |
|---|---|---|
No discrete prompt usage | 65.78±3.23 | 65.83±3.01 |
No filtering | 65.12±2.52 | 64.93±2.88 |
Euclidean distance | 70.75±2.81 | 70.23±2.33 |
Cosine similarity | 72.22±2.10 | 72.57±2.3 |
Pearson correlation coefficient | 68.32±2.05 | 71.57±2.49 |
